WEBVTT

1
00:00:00.291 --> 00:00:02.652
But in the Navier Stokes, and that's why it's so interesting.

2
00:00:03.333 --> 00:00:08.456
Because the Navier Stokes has this non-linearity, but it also has this smoothing term.

3
00:00:08.536 --> 00:00:10.357
It has this counterforce, which is the viscosity.

4
00:00:10.397 --> 00:00:12.278
And so you're asking, who's going to win?

5
00:00:12.638 --> 00:00:14.379
And are they always going to win?

6
00:00:14.759 --> 00:00:17.720
There are questions about credit, commercial incentives, accountability.

7
00:00:18.081 --> 00:00:18.281
Yes.

8
00:00:19.609 --> 00:00:26.738
But I do believe that the risks deserve real attention without being mired in those other things.

9
00:00:26.898 --> 00:00:27.139
Yeah.

10
00:00:27.499 --> 00:00:29.642
Those other things can be true and the risk can still be real.

11
00:00:30.263 --> 00:00:31.945
Okay, so just turn it off.

12
00:00:32.165 --> 00:00:32.285
Yeah.

13
00:00:32.305 --> 00:00:34.027
What does just turn it off actually mean?

14
00:00:34.047 --> 00:00:34.508
Yeah.

15
00:00:34.668 --> 00:00:35.389
Hello, Internet.

16
00:00:35.409 --> 00:00:43.454
This is your captain speaking, Lester Nare, joined as always by my co-host and our resident PhD, Krishna Chowdhury.

17
00:00:43.914 --> 00:00:51.180
We are about to get into what will be our most requested episode in the history of From First Principles.

18
00:00:51.780 --> 00:00:56.263
OpenAI announced a solution to one of the Millennium Problems.

19
00:00:56.343 --> 00:01:01.447
Many of you have either seen that announcement or all of the fallout and drama problems.

20
00:01:01.987 --> 00:01:05.828
as it relates to AI that has arisen since that announcement.

21
00:01:05.908 --> 00:01:11.309
And so a quick overview of how we're going to structure this deep dive episode.

22
00:01:11.609 --> 00:01:13.570
We'll be lucky to get it under three hours.

23
00:01:13.930 --> 00:01:18.891
So we're going to do something a little bit different than some of the mainstream coverage that starts with the drama.

24
00:01:19.671 --> 00:01:26.053
First, we're going to ground ourselves in the fundamental understanding of the mathematics as it relates to the OpenAI solution.

25
00:01:26.473 --> 00:01:31.314
What are the Navier-Stokes equations and what is the existence and smoothness problem?

26
00:01:32.394 --> 00:01:34.836
And how exactly did they solve it?

27
00:01:35.456 --> 00:01:47.603
And then we will obviously move into some of the drama and tragedy that has arisen as a result of this and other progression in AI over the summer into the early fall.

28
00:01:48.915 --> 00:01:51.157
This is going to be a nuanced episode.

29
00:01:51.297 --> 00:01:54.359
We're going to talk about the science and some of the larger complexities.

30
00:01:54.759 --> 00:01:58.742
If you want to understand science headlines, be sure to subscribe.

31
00:01:58.962 --> 00:02:05.927
As always, we are going to talk about the science from the ground up today because this is From First Principles.

32
00:02:20.305 --> 00:02:22.326
It's not a question of if, but when.

33
00:02:23.627 --> 00:02:29.190
All of the people that have been paying attention to what AI has been doing in mathematics, that is what they are asking.

34
00:02:29.931 --> 00:02:35.014
When is AI going to solve one of the six open millennium problems?

35
00:02:35.294 --> 00:02:36.675
There's six of them that are left.

36
00:02:36.775 --> 00:02:38.296
They're kind of like the infinity stones.

37
00:02:38.996 --> 00:02:41.938
And I'd say AI is kind of like Thanos, you know?

38
00:02:42.678 --> 00:02:43.439
AI is coming.

39
00:02:44.219 --> 00:02:46.582
And the question was always, when is he going to get the first stone?

40
00:02:46.642 --> 00:02:48.504
Well, it seems he might have got it.

41
00:02:48.824 --> 00:02:49.064
Okay.

42
00:02:49.405 --> 00:02:53.629
The Navier Stokes existence and smoothness problem might have been solved.

43
00:02:54.089 --> 00:02:55.651
And naturally we were on vacation.

44
00:02:55.771 --> 00:02:58.254
I was in beautiful Lake Geneva in Wisconsin.

45
00:02:58.714 --> 00:02:59.255
Yeah.

46
00:02:59.715 --> 00:03:02.957
And somebody forwarded me a tweet and this is what I was feeling.

47
00:03:02.977 --> 00:03:03.918
I made a meme about it.

48
00:03:04.378 --> 00:03:06.880
You know, we're, we're also anticipating GTA six.

49
00:03:06.960 --> 00:03:07.220
Yes.

50
00:03:07.480 --> 00:03:13.564
And this is the, ah, here we go again, because honestly, I didn't, I didn't want to keep talking about AI and math.

51
00:03:13.604 --> 00:03:15.265
We'd already done the Jacobian conjecture.

52
00:03:15.305 --> 00:03:22.289
We'd already done the Riemann hypothesis or a problem related to that, both by anthropic, but this, this, this is history.

53
00:03:22.609 --> 00:03:24.911
If it's true, you know,

54
00:03:26.370 --> 00:03:27.591
This is a big deal.

55
00:03:28.051 --> 00:03:29.392
And it's not anthropic this time.

56
00:03:29.732 --> 00:03:30.493
It's open AI.

57
00:03:31.453 --> 00:03:36.797
And the announcement was met with a different kind of reaction than the other two.

58
00:03:37.217 --> 00:03:43.401
In the other two, there was a reaction of, oh my gosh, AI is so good at mathematics.

59
00:03:43.461 --> 00:03:44.322
This is incredible.

60
00:03:45.002 --> 00:03:48.524
There's a bit of an existential crisis in mathematics and so on and so forth.

61
00:03:48.564 --> 00:03:51.146
But here, there was also something else.

62
00:03:51.186 --> 00:03:51.947
There was a fight.

63
00:03:52.787 --> 00:03:56.989
There were allegations of academic misconduct.

64
00:03:57.149 --> 00:03:59.210
There were questions about who deserved credit.

65
00:03:59.731 --> 00:04:05.814
There's questions about what OpenAI proved themselves and what they maybe stole from other people's work.

66
00:04:05.894 --> 00:04:17.600
That's someone else being the NYU mathematician Tristan Buckmaster and anthropic researcher Levant Alpoghe, who was the hero behind the Jacobian conjecture a few episodes back.

67
00:04:18.349 --> 00:04:19.550
So there's a lot to unpack.

68
00:04:20.031 --> 00:04:23.654
But before we get into any of that, as you said, we're going to do something very deliberately.

69
00:04:23.674 --> 00:04:25.176
We're going to talk about the mathematics.

70
00:04:25.616 --> 00:04:38.229
Because at the end of this two-page document that Buckmaster published, which really kicked off the controversy, this document where he talked about all the shenanigans that had happened behind the scenes, he said,

71
00:04:38.389 --> 00:04:43.270
He makes the argument that if OpenAI has really solved Navier-Stokes, then they should just say so.

72
00:04:43.851 --> 00:04:46.791
And he'd rather be discussing the mathematics than anything else.

73
00:04:47.451 --> 00:04:48.332
And honestly, same.

74
00:04:48.432 --> 00:04:49.172
For me, same.

75
00:04:49.632 --> 00:04:51.012
So that's where we're going to begin.

76
00:04:51.032 --> 00:04:52.153
We're going to do the math first.

77
00:04:52.173 --> 00:04:53.473
We're going to do the drama second.

78
00:04:53.873 --> 00:04:55.594
So here's the roadmap for the episode.

79
00:04:55.914 --> 00:05:00.815
We're going to start from first principles and try to build up the Navier-Stokes equations.

80
00:05:01.175 --> 00:05:02.456
They're a set of two equations.

81
00:05:03.216 --> 00:05:07.497
And at the end of that segment, I want you to have an intuitive understanding of every single term.

82
00:05:08.177 --> 00:05:08.917
in that equation.

83
00:05:08.957 --> 00:05:11.278
So it's not just a bunch of words and fancy symbols.

84
00:05:11.919 --> 00:05:18.061
At the end of that, we'll confront the actual millennium problem, this question of existence and smoothness.

85
00:05:18.461 --> 00:05:20.102
There is going to be some calculus involved.

86
00:05:20.122 --> 00:05:22.823
There's going to be some equations, but I have a lot of visuals.

87
00:05:23.203 --> 00:05:24.624
So hopefully it's not going to be that bad.

88
00:05:24.684 --> 00:05:26.845
And we'll describe it for our audio listeners.

89
00:05:26.905 --> 00:05:31.947
But I do suggest this is a very visual heavy episode, as most of our episodes are.

90
00:05:33.067 --> 00:05:42.791
Then we'll take a tour through the mathematics over the last several decades and get to the main event, which is OpenAI's 165-page proof.

91
00:05:42.991 --> 00:05:44.452
The claim is that they have solved it.

92
00:05:44.572 --> 00:05:47.073
Now, I am not a fluid dynamicist.

93
00:05:48.093 --> 00:05:54.918
And I'm certainly not an expert in the kinds of linear partial differential equations that the Navier-Stokes problem is.

94
00:05:55.318 --> 00:06:02.083
This is a problem that has attracted some of the greatest mathematicians of our time, including Terence Tao, among others.

95
00:06:02.263 --> 00:06:07.447
So I'm approaching this from a perspective of a physicist who's trying to understand fluid mechanics.

96
00:06:07.787 --> 00:06:15.233
And so I went back to our trusty old Landau-Lefschitz course in theoretical physics, volume six, Fluid Mechanics.

97
00:06:15.573 --> 00:06:19.735
Landau-Lifschitz goes through some amazing, amazing mathematics here.

98
00:06:19.755 --> 00:06:26.298
And Navier-Stokes is in the early chapters because there's so much more to fluids than just Navier-Stokes.

99
00:06:26.418 --> 00:06:29.940
Shout out to my professor, Robin Bruinsma, who taught me this course.

100
00:06:30.060 --> 00:06:34.482
It was an amazing course where we got to apply some of my favorite mathematics, i.e.

101
00:06:34.502 --> 00:06:40.125
complex analysis, to calculate the lift on a wing of an airplane.

102
00:06:40.525 --> 00:06:53.124
using a complex integral, which was one of the most mind-blowing things that you could use complex mechanics to talk about real things, not just prime numbers, as we've been known.

103
00:06:53.584 --> 00:06:55.086
So at the end of all that, we're going to be

104
00:06:56.428 --> 00:07:03.471
understanding the solution, and then we'll talk about the drama, and then all of the existential stuff about what all of this means.

105
00:07:03.531 --> 00:07:04.811
Because we're going to have to zoom out.

106
00:07:04.851 --> 00:07:06.292
This is not happening in a vacuum.

107
00:07:06.672 --> 00:07:11.674
And there is a lot of crazy developments that are honestly happening daily.

108
00:07:12.334 --> 00:07:13.755
It is really happening daily.

109
00:07:13.955 --> 00:07:21.878
And once Krishna has kind of walked us through the mathematics, I kind of want to look at the larger world that this announcement is,

110
00:07:22.498 --> 00:07:23.059
happened in.

111
00:07:24.420 --> 00:07:30.944
There is legitimate disputes over the credit and how we think about acknowledgement in scientific discovery and understanding.

112
00:07:31.985 --> 00:07:41.212
And this is also happening in the context of an already existing and accelerating battle between multiple stakeholders as it relates to this issue.

113
00:07:41.232 --> 00:07:45.515
You have the frontier labs, you have the investor and venture capital community

114
00:07:46.015 --> 00:07:49.079
You have politicians beginning to finally weigh in on this.

115
00:07:49.619 --> 00:07:57.348
Obviously, you have the general public who, for many people, don't really see real-world impact on this in their everyday life.

116
00:07:58.009 --> 00:08:05.677
And for context around some of these AI tools, I personally have been a user of them lately.

117
00:08:06.293 --> 00:08:11.974
just as a technologist and someone who's interested in these subjects in the way that we cover on the show, since GPT-2.

118
00:08:12.534 --> 00:08:19.356
And I've seen what the progress looks like from a very direct personal context.

119
00:08:19.956 --> 00:08:32.919
And I do think that there is a challenge in discussing this issue right now because you have to fit into one of two boxes, which is this is all hype and this is going to cause human extinction.

120
00:08:33.039 --> 00:08:33.139
Yeah.

121
00:08:33.159 --> 00:08:33.299
Yeah.

122
00:08:33.619 --> 00:08:41.401
And in my view, there's a lot of space in between this is all hype and this is going to cause human extinction.

123
00:08:41.601 --> 00:08:45.463
And I would love for us to be better able to have conversations within that gap.

124
00:08:45.823 --> 00:08:49.584
And so we're going to talk through some of the larger context.

125
00:08:50.104 --> 00:08:58.552
to help frame this because this will be a technology, a transformation that will be with us for the foreseeable future.

126
00:08:58.972 --> 00:09:01.715
And I think it's going to be important to look at these two things.

127
00:09:01.735 --> 00:09:12.725
And we'll look at some concrete examples of other things that have happened, like the hugging face incident, where there were some issues with open AI's models as it relates to the company hugging face, because I think that'll help,

128
00:09:13.345 --> 00:09:21.253
Really ground us around this shift from this is fancy autocomplete or this is just a better way to do search.

129
00:09:22.113 --> 00:09:28.660
And this agentic world that has been actually a driver behind how OpenAI works.

130
00:09:29.320 --> 00:09:32.002
got to this solution.

131
00:09:32.082 --> 00:09:33.943
So we're going to get into all of that.

132
00:09:33.963 --> 00:09:36.045
A real brief moment of housekeeping.

133
00:09:36.285 --> 00:09:41.448
For those of you who may be joining us for the first time, welcome to the best science show on the planet.

134
00:09:41.468 --> 00:09:46.872
You can find us on all platforms where you listen to your favorite podcasts, as well as social for clips.

135
00:09:47.472 --> 00:09:56.923
Our website has all of the research papers, background material, and other ways to look into what's happening in Frontier Science Research at ffppod.com.

136
00:09:57.424 --> 00:10:05.253
Any engagement that you do on our videos on all platforms help us reach more people with what we think are some of the most fun and fun things.

137
00:10:05.533 --> 00:10:08.398
and informative conversations about science.

138
00:10:08.859 --> 00:10:11.063
And we have some great merch.

139
00:10:11.263 --> 00:10:17.694
So with that, we're going to jump right in to what is the meat and bones of what's going to be, I think, a very fantastic episode.

140
00:10:18.020 --> 00:10:20.261
Yes, and we will start with the mathematics.

141
00:10:20.821 --> 00:10:23.122
So these are the Navier-Stokes equations.

142
00:10:23.642 --> 00:10:24.302
There's two of them.

143
00:10:24.902 --> 00:10:26.923
And our goal is to understand these equations.

144
00:10:26.963 --> 00:10:31.164
These equations are an equation of motion that describes fluids.

145
00:10:31.584 --> 00:10:35.165
How fluids behave, what are the forces that they exert.

146
00:10:35.725 --> 00:10:41.267
If I know what the fluid is doing right now, can I predict what the fluid will be doing at some later time?

147
00:10:41.807 --> 00:10:44.288
These are the equations that allow that to happen.

148
00:10:45.028 --> 00:10:48.031
And when I say they're an equation of motion, that's what I mean.

149
00:10:48.071 --> 00:10:53.517
They're like, they're telling you how the equation, how the fluid is going to move around in some space.

150
00:10:53.917 --> 00:11:01.205
Now, in classical systems, the starting point for all equations of motion is always Newton's laws, okay?

151
00:11:02.718 --> 00:11:10.162
Newton's first law is an object at rest stays at rest, and an object in motion will keep going unless it's acted on by another force.

152
00:11:10.663 --> 00:11:16.006
So, you know, if you're in empty space and you just have a satellite that's moving in one direction, it's going to keep moving.

153
00:11:16.426 --> 00:11:17.647
What happens when there's a force?

154
00:11:17.727 --> 00:11:20.669
Well, when there's a force, there's an acceleration.

155
00:11:20.769 --> 00:11:22.590
It's not a velocity.

156
00:11:23.270 --> 00:11:25.852
It's not like you need a force to keep it going.

157
00:11:26.272 --> 00:11:27.653
If it's going, it'll just keep going.

158
00:11:27.673 --> 00:11:28.093
Okay.

159
00:11:28.355 --> 00:11:30.936
If there's a force, it'll change how it's moving.

160
00:11:31.016 --> 00:11:32.417
So that's called an acceleration.

161
00:11:32.757 --> 00:11:35.378
It's a change in velocity.

162
00:11:35.798 --> 00:11:43.702
And the relationship between the amount of force that I put on an object and how fast it's going to accelerate is dependent on the mass.

163
00:11:44.944 --> 00:11:46.145
Mass is a kind of inertia.

164
00:11:46.165 --> 00:11:52.190
It's like, how much resistance to a force do I have, is literally what mass is.

165
00:11:52.490 --> 00:12:05.960
And there's a pretty common demonstration that you see in physics classrooms in high school, where you've got two rubber bands, they're the same rubber band, but one of them is attached to a block of wood, and the other one is attached to two of the same blocks of wood.

166
00:12:06.300 --> 00:12:10.663
And you can see that the single block of wood accelerates a lot faster than...

167
00:12:10.883 --> 00:12:12.064
than the one with two blocks of wood.

168
00:12:12.084 --> 00:12:16.645
That's because the two blocks of wood has twice the mass, so it's going to have half the acceleration.

169
00:12:16.945 --> 00:12:17.806
There's Newton's second law.

170
00:12:18.086 --> 00:12:25.349
And Newton's third law is something about conservation of momentum, which is like, you know, if I'm pushing on an object, then that object pushes back on me.

171
00:12:26.049 --> 00:12:27.250
So those are the three laws.

172
00:12:27.410 --> 00:12:35.693
And if we want to understand how something moves around, all we have to do is figure out what is the nature of the forces on this object and,

173
00:12:35.813 --> 00:12:41.397
And if we're good enough at the math, we can figure out how this object is going to move forever.

174
00:12:42.157 --> 00:12:45.440
This is partly what enables us, for example, to land on the moon.

175
00:12:45.680 --> 00:12:46.040
Everything.

176
00:12:46.100 --> 00:12:50.263
Or Mars, all of these kinds of, obviously, much more surface area than that.

177
00:12:50.303 --> 00:12:53.585
But this is why we can make predictions over such long periods of time.

178
00:12:53.665 --> 00:12:57.307
Yeah, really, in order to get to the moon, you just need Newton's laws.

179
00:12:57.627 --> 00:13:00.788
I guess you need electricity to work the capsule or whatever.

180
00:13:01.248 --> 00:13:08.752
But if there was some analog version where you had fluids that did computation or something, you could land on the moon with just Newton's laws.

181
00:13:09.192 --> 00:13:11.393
So let's take, for example, the mass on a spring.

182
00:13:11.693 --> 00:13:24.877
I want to go through a singular example of a very simple problem where I can have an equation of motion derived from Newton's laws, and I can then figure out what that thing is going to do forever.

183
00:13:25.257 --> 00:13:27.718
A common example is a mass on a spring.

184
00:13:27.978 --> 00:13:34.260
A spring is a substance that obeys Hooke's law, meaning if I pull on the thing, it's going to resist and it's going to pull back.

185
00:13:34.760 --> 00:13:38.001
And how much it's going to pull back is dependent on how much I'm stretching it.

186
00:13:38.541 --> 00:13:43.924
If I stretch it twice as far, the pullback, the force backwards, is twice as much.

187
00:13:44.224 --> 00:13:45.144
That's called Hooke's Law.

188
00:13:45.484 --> 00:13:49.767
And you usually see it as the force is equal to negative kx.

189
00:13:50.207 --> 00:13:51.767
k is how stiff the spring is.

190
00:13:52.628 --> 00:13:54.049
x is how far I've pulled.

191
00:13:54.409 --> 00:13:57.030
And so, you know, the larger I pull, the more it pulls back.

192
00:13:57.290 --> 00:14:02.293
And the negative is because it's pulling in the opposite direction of how I'm distorting the spring.

193
00:14:02.473 --> 00:14:02.713
Okay?

194
00:14:02.833 --> 00:14:03.113
Mm-hmm.

195
00:14:03.613 --> 00:14:05.635
So that's Hooke's Law, f equals negative kx.

196
00:14:05.995 --> 00:14:10.058
Now, force, according to Newton's second law, is mass times acceleration, right?

197
00:14:10.418 --> 00:14:16.282
And the acceleration is really a second derivative of the position, of the x, right?

198
00:14:16.402 --> 00:14:22.707
The acceleration is a change in velocity, and the velocity is a change in position with respect to time.

199
00:14:23.007 --> 00:14:30.032
So it's really a double change, which is why I have mass times a second derivative in position is equal to negative kx.

200
00:14:30.092 --> 00:14:31.653
Notice, this is an equation of motion.

201
00:14:31.673 --> 00:14:32.013
Mm-hmm.

202
00:14:32.933 --> 00:14:37.620
I've got a single variable that I need to solve for, which is x, the position.

203
00:14:38.201 --> 00:14:42.147
And the key thing is, the force is dependent on the position.

204
00:14:43.447 --> 00:14:47.230
But the force is related to how fast the position accelerates.

205
00:14:47.670 --> 00:14:54.435
So now I've got an equation that is in a single variable, x, and it's a differential equation with respect to time.

206
00:14:54.835 --> 00:14:57.036
This is something I can solve.

207
00:14:57.316 --> 00:14:58.737
And that's what most people do.

208
00:14:58.917 --> 00:14:59.958
You can solve this thing.

209
00:15:00.138 --> 00:15:01.219
It's a differential equation.

210
00:15:01.899 --> 00:15:05.462
And if you solve it, you get a very nice solution.

211
00:15:05.482 --> 00:15:05.902
Right?

212
00:15:06.218 --> 00:15:08.659
you get a very nice solution that is an oscillation.

213
00:15:09.039 --> 00:15:10.939
And I think we've got a little video of that.

214
00:15:10.959 --> 00:15:11.700
This is what happens.

215
00:15:12.140 --> 00:15:21.883
You take your equation of motion and you say, what is something where if I take a second derivative, I take a derivative once and I take the derivative again, I get the negative of the thing back.

216
00:15:22.583 --> 00:15:23.744
And that's cosine and sine.

217
00:15:23.984 --> 00:15:26.326
The derivative of cosine is negative sine.

218
00:15:26.566 --> 00:15:28.467
The derivative of sine is cosine.

219
00:15:28.487 --> 00:15:31.689
So I'd go from cosine to negative sine to negative cosine.

220
00:15:31.769 --> 00:15:33.651
Well, that's just the negative of what I started with.

221
00:15:34.151 --> 00:15:36.032
And there, I have an oscillatory motion.

222
00:15:36.473 --> 00:15:40.175
So this is the paradigm of how we do physics in general.

223
00:15:40.535 --> 00:15:41.496
We have Newton's laws.

224
00:15:42.177 --> 00:15:43.377
We analyze our system.

225
00:15:43.758 --> 00:15:47.881
And we figure out what is the differential equation that the system obeys.

226
00:15:48.641 --> 00:15:49.502
And then we solve this.

227
00:15:49.542 --> 00:15:56.166
And if we're lucky enough, the differential equation is simple enough that the mathematicians have figured out how to solve this differential equation.

228
00:15:56.246 --> 00:15:58.948
And the physicists can then borrow that intuition.

229
00:15:59.569 --> 00:16:06.514
And boom, you've got a solution that tells you, for any time, I can plug in the time and I can tell you where the mass is going to be.

230
00:16:07.334 --> 00:16:16.600
And it's a way in which we can now simulate and create replications such that we can predict how a system will act under certain circumstances.

231
00:16:16.660 --> 00:16:17.580
Yes, exactly.

232
00:16:18.021 --> 00:16:20.822
And you can make things even more complicated.

233
00:16:21.183 --> 00:16:24.665
For example, you can take the same spring and you can add friction.

234
00:16:25.185 --> 00:16:27.266
In this case, there's like some kind of damping force.

235
00:16:28.647 --> 00:16:31.049
And what that means is that in your differential equation...

236
00:16:31.812 --> 00:16:33.774
you've got a term that's related to velocity.

237
00:16:33.954 --> 00:16:39.518
Because the faster I move through some... Let's say I dunk this thing in honey.

238
00:16:39.779 --> 00:16:44.883
The faster I move through this thing, the more the honey is going to resist my motion.

239
00:16:45.143 --> 00:16:47.085
So that is dependent on the first derivative.

240
00:16:48.165 --> 00:16:49.565
of position, just the velocity.

241
00:16:49.926 --> 00:16:54.507
And if I add that term, then I get a new solution, which is the damped harmonic oscillator.

242
00:16:54.747 --> 00:16:56.227
And that's the red that you see over there.

243
00:16:56.287 --> 00:17:00.608
So in the undamped case, you've got an oscillation that just bounces over and over and over again.

244
00:17:01.168 --> 00:17:04.349
And then in the damped case, you've got an oscillation that dies out very quickly.

245
00:17:04.589 --> 00:17:17.172
For those of you who are listening to us in the car, you can thank damped harmonic motion for your suspension because there's a particular value of the damping that makes all these oscillations go to zero very fast.

246
00:17:17.672 --> 00:17:18.852
It's called critical damping.

247
00:17:19.252 --> 00:17:28.294
And a lot of times car manufacturers want their springs, the suspension in their springs, to have critically damped so that you get bounced on a... What are those called?

248
00:17:28.314 --> 00:17:28.694
The holes?

249
00:17:28.974 --> 00:17:29.594
The potholes?

250
00:17:30.174 --> 00:17:33.215
You bounce on a pothole, but you don't bounce up and down forever.

251
00:17:33.235 --> 00:17:35.215
You just immediately bounce and then you go right back.

252
00:17:35.595 --> 00:17:37.316
And there's no bouncing whatsoever.

253
00:17:37.336 --> 00:17:39.436
And that's because the spring is critically damped.

254
00:17:39.496 --> 00:17:42.516
That has to do with how much damping is happening.

255
00:17:43.117 --> 00:17:44.797
My point is you can make the...

256
00:17:45.618 --> 00:17:49.229
equations of motion a bit more complicated, and still be able to solve stuff.

257
00:17:49.622 --> 00:17:55.184
This is the visual I think of is seeing those meme videos of those giant trucks with giant suspension.

258
00:17:55.204 --> 00:18:00.465
That's visible going over, you know, this hump dirt, you know, dirt track.

259
00:18:00.965 --> 00:18:02.046
And they're chilling.

260
00:18:02.206 --> 00:18:02.426
Yeah.

261
00:18:02.466 --> 00:18:02.646
Yeah.

262
00:18:02.666 --> 00:18:02.846
Yeah.

263
00:18:03.066 --> 00:18:09.168
There's all these like Chinese propaganda videos too, where they show like the luxury cars of China versus the luxury cars of Germany.

264
00:18:09.388 --> 00:18:13.149
And it's like the luxury cars of China are just like, it's, they're like on a cloud.

265
00:18:13.509 --> 00:18:17.530
And then the ones like the BMW is like, it's like, okay, well, is that a new BMW or like,

266
00:18:17.830 --> 00:18:20.192
You never know with those propaganda videos.

267
00:18:20.272 --> 00:18:21.473
Or is it just straight up AI?

268
00:18:21.933 --> 00:18:25.415
Anyways, for every case, the logic has been pretty simple.

269
00:18:26.016 --> 00:18:27.657
The point is we start from Newton's laws.

270
00:18:27.697 --> 00:18:29.038
We understand what the forces are.

271
00:18:29.138 --> 00:18:31.299
And then we try to set up the equations of motion.

272
00:18:31.579 --> 00:18:32.640
This works for springs.

273
00:18:32.980 --> 00:18:34.121
This works for planets.

274
00:18:34.761 --> 00:18:39.945
For a planet, the force is proportional to 1 over r squared, how far away my thing is.

275
00:18:40.233 --> 00:18:43.674
and it's pointing directly at the central body.

276
00:18:44.075 --> 00:18:47.156
And again, that's a force that is a function of position.

277
00:18:47.756 --> 00:18:53.398
And now I can solve for what the position of the planet is because I can set up a differential equation that does so.

278
00:18:53.538 --> 00:18:59.220
Okay, I can set up a differential equation for a three body problem, but crucially I cannot solve that.

279
00:18:59.521 --> 00:19:02.142
So I can set up the equation of motion, but I don't have a solution for it.

280
00:19:02.602 --> 00:19:04.202
So there's caveats, right?

281
00:19:04.583 --> 00:19:07.704
Okay, so now let's try to understand fluids.

282
00:19:08.464 --> 00:19:15.187
I see the setup we're going with here because we're saying we can understand the world around us and make predictions using math.

283
00:19:15.587 --> 00:19:16.428
Yeah, and Newton's laws.

284
00:19:16.508 --> 00:19:19.249
And Newton's laws for certain physical systems.

285
00:19:19.269 --> 00:19:19.489
Yes.

286
00:19:20.069 --> 00:19:24.912
And we just defined examples of those physical systems in which we can do so.

287
00:19:25.032 --> 00:19:25.232
Yes.

288
00:19:27.368 --> 00:19:35.933
Now, I think where we're going is fluids maybe have some unique aspects that make that different.

289
00:19:36.194 --> 00:19:38.035
Yes, it makes it different and difficult.

290
00:19:38.215 --> 00:19:38.475
Okay.

291
00:19:38.495 --> 00:19:38.735
Okay?

292
00:19:39.436 --> 00:19:40.556
It's different in the...

293
00:19:40.836 --> 00:19:46.120
So, first of all, it's same, same a little bit because each part of the fluid still obeys Newton's laws.

294
00:19:46.320 --> 00:19:46.500
Yes.

295
00:19:46.560 --> 00:19:47.921
If I were to look at a single...

296
00:19:48.361 --> 00:19:52.482
particle of fluid, let's say I drop a dust grain into a moving river, right?

297
00:19:52.742 --> 00:19:57.683
That dust grain is going to experience forces because water molecules are pushing on the thing.

298
00:19:58.303 --> 00:20:00.643
And those forces are going to make it accelerate.

299
00:20:01.343 --> 00:20:02.924
Okay, so Newton's law is still a work.

300
00:20:03.244 --> 00:20:03.724
100%.

301
00:20:03.764 --> 00:20:03.964
Right?

302
00:20:04.764 --> 00:20:14.266
The problem, though, is that Newton's laws are tenable and they're, like, tractable if it's a single thing that's, like, moving around.

303
00:20:15.310 --> 00:20:20.995
In a fluid, for example, even in this cup, right, of water, I have 10 to the 3 molecules.

304
00:20:21.195 --> 00:20:22.396
No, no, no, it's not 10 to the 3.

305
00:20:22.436 --> 00:20:23.196
That's only 1,000.

306
00:20:23.597 --> 00:20:26.839
10 to the 23 molecules of water moving around.

307
00:20:27.220 --> 00:20:27.460
Okay?

308
00:20:27.700 --> 00:20:28.020
So what?

309
00:20:28.060 --> 00:20:33.405
Do you expect me to apply Newton's laws to every single molecule that's in this water?

310
00:20:33.785 --> 00:20:35.386
It doesn't make any sense.

311
00:20:35.747 --> 00:20:36.988
Sure, that's one way of doing it.

312
00:20:37.028 --> 00:20:39.450
If I had infinite compute, which I don't.

313
00:20:40.502 --> 00:20:50.066
I'd like to understand this thing at a level of granularity where I can make predictions, but maybe I don't have to worry about every single molecule and what it is doing.

314
00:20:50.186 --> 00:20:54.908
We don't want to have to simulate everything because the system is so large as compared to what we were doing before.

315
00:20:54.988 --> 00:20:58.189
Yeah, the system is large and it's self-interacting is the other point, right?

316
00:20:58.209 --> 00:21:00.930
The water molecules are interacting with other water molecules.

317
00:21:01.330 --> 00:21:04.071
And like, it's just a whole mess, right?

318
00:21:04.431 --> 00:21:07.032
So it's not a new kind of physics.

319
00:21:07.172 --> 00:21:08.373
It's the same physics, but it's...

320
00:21:09.013 --> 00:21:12.035
being applied everywhere in the system.

321
00:21:12.275 --> 00:21:13.236
Everywhere all at once.

322
00:21:13.356 --> 00:21:14.076
And all at once.

323
00:21:14.317 --> 00:21:14.777
Exactly.

324
00:21:15.037 --> 00:21:16.618
And so we need a new language.

325
00:21:16.978 --> 00:21:17.339
Okay.

326
00:21:17.519 --> 00:21:18.960
And perhaps here's the point.

327
00:21:19.908 --> 00:21:22.710
What do we want to actually describe about this fluid?

328
00:21:22.990 --> 00:21:26.033
We don't want to describe the individual molecules, as I just said.

329
00:21:26.333 --> 00:21:33.458
It would be crazy to try and describe the position of every single molecule as a function of time, like we could with that mass on a spring, right?

330
00:21:34.779 --> 00:21:35.980
That doesn't really make sense here.

331
00:21:36.020 --> 00:21:44.967
So instead of tracking individual molecules, we are going to track the velocity of the fluid itself, okay, at different spots.

332
00:21:45.147 --> 00:21:45.387
Okay.

333
00:21:45.760 --> 00:21:50.066
And it turns out this has all of the necessary information that we need.

334
00:21:50.747 --> 00:21:51.248
Interesting.

335
00:21:51.268 --> 00:21:51.488
Okay?

336
00:21:51.729 --> 00:21:52.490
This is what we're doing.

337
00:21:52.570 --> 00:21:58.058
On the left-hand side, you see a video of sort of dust particles, let's say, that are in a chunk of a river.

338
00:21:58.418 --> 00:21:58.678
Okay?

339
00:21:58.719 --> 00:22:01.402
So we can see the dust particles that are moving from left to right.

340
00:22:01.422 --> 00:22:01.603
Mm-hmm.

341
00:22:02.298 --> 00:22:05.041
Now, I don't want to worry about all of the dust particles.

342
00:22:05.081 --> 00:22:12.008
So instead, what I worry about is what is the velocity of the particles at that point.

343
00:22:12.809 --> 00:22:18.334
And those can be represented by little arrows that just represent where the dust particles are going at that point.

344
00:22:18.875 --> 00:22:20.697
Now, crucially, the dust particles are moving.

345
00:22:21.337 --> 00:22:25.682
But this velocity field that I've made of a bunch of arrows that are representing particles,

346
00:22:26.058 --> 00:22:28.301
all of the velocities, that thing is stationary.

347
00:22:28.881 --> 00:22:29.222
You see?

348
00:22:29.742 --> 00:22:32.926
Because it just represents, this is the direction in which I'm moving.

349
00:22:33.326 --> 00:22:33.546
Okay?

350
00:22:33.987 --> 00:22:38.312
This is the central element that we are worried about.

351
00:22:38.412 --> 00:22:38.712
Okay.

352
00:22:38.852 --> 00:22:39.113
Okay?

353
00:22:39.593 --> 00:22:41.015
This is a velocity field.

354
00:22:41.636 --> 00:22:46.441
It's a field because it has a value at every point in space and time.

355
00:22:47.705 --> 00:22:49.265
and it's a velocity, so it's got a vector.

356
00:22:49.326 --> 00:22:50.726
It's a vector field, okay?

357
00:22:51.166 --> 00:22:53.546
And usually this thing is represented as a U.

358
00:22:53.847 --> 00:22:55.387
We don't want to use V for some reason.

359
00:22:55.407 --> 00:22:56.767
I think it's a historical reason.

360
00:22:57.147 --> 00:23:02.909
So in every case, the U that we're going to see later on in this episode, that is a velocity field.

361
00:23:03.109 --> 00:23:07.790
It's a vector field everywhere in space, and it represents what the velocity is at that particular moment.

362
00:23:07.830 --> 00:23:14.271
So if I have a velocity vector that's pointed in this direction, that means water is moving past in this direction.

363
00:23:14.611 --> 00:23:15.872
If, on the other hand, it's over here,

364
00:23:17.493 --> 00:23:20.217
towards that arrow, towards where the arrow is pointing.

365
00:23:20.237 --> 00:23:23.563
And the component parts that make this velocity field are...

366
00:23:24.640 --> 00:23:28.021
Some position in 3D space, which is our X, Y, and Z. Yeah.

367
00:23:28.101 --> 00:23:33.262
So we know like where in this cube, the velocity field that we're specifically referring to exists.

368
00:23:33.742 --> 00:23:36.823
And then time allows us to create the arrow of time.

369
00:23:36.923 --> 00:23:37.163
Yeah.

370
00:23:37.343 --> 00:23:38.844
Of the direct, like at that point.

371
00:23:39.084 --> 00:23:40.704
At that point, it's moving in this direction.

372
00:23:40.764 --> 00:23:40.984
Yeah.

373
00:23:41.124 --> 00:23:41.324
Yeah.

374
00:23:41.344 --> 00:23:41.484
Right.

375
00:23:41.724 --> 00:23:43.025
And it could change, right?

376
00:23:43.105 --> 00:23:46.566
Maybe I introduce the flow and then it starts moving in some other direction.

377
00:23:46.726 --> 00:23:46.846
Right.

378
00:23:46.866 --> 00:23:48.286
And so the velocity field can change.

379
00:23:49.287 --> 00:23:53.350
This is a very simple case, right, of just like particles, water moving from left to right.

380
00:23:53.650 --> 00:23:55.992
There can be other more complicated cases.

381
00:23:56.313 --> 00:23:59.275
For example, this is the velocity field of a tornado.

382
00:23:59.936 --> 00:24:05.000
If I were to look like up the axis of a tornado, it's rotating around.

383
00:24:05.481 --> 00:24:12.987
And so imagine if you're looking like up through the eye of the tornado, you would see particles and clouds and everything moving around.

384
00:24:13.710 --> 00:24:14.571
around you, right?

385
00:24:14.611 --> 00:24:16.033
Like in a circle above you.

386
00:24:16.454 --> 00:24:19.818
And the velocity field for that looks like a bunch of arrows that are

387
00:24:20.980 --> 00:24:22.180
sort of circulating around.

388
00:24:22.200 --> 00:24:23.260
Okay?

389
00:24:24.181 --> 00:24:28.141
And the color of these arrows represents how strong the velocity is.

390
00:24:28.581 --> 00:24:28.822
Okay?

391
00:24:28.902 --> 00:24:39.043
So the vector is big near the eye, and as you move farther and farther away from the axis of the tornado, the vector field diminishes in magnitude because things are moving slower.

392
00:24:39.384 --> 00:24:40.084
This makes sense.

393
00:24:40.104 --> 00:24:46.625
This is why when you have a tornado or a hurricane approaching you, if you live on the coast, the outer parts, you can feel it coming.

394
00:24:46.745 --> 00:24:46.885
Yeah.

395
00:24:46.925 --> 00:24:47.025
Yeah.

396
00:24:47.325 --> 00:24:49.766
Because the outer parts are a little bit slower than the inner parts.

397
00:24:49.786 --> 00:24:53.708
Yeah, when you get to the inner parts, then it's like, okay, then the palm tree is like... On the side.

398
00:24:53.728 --> 00:24:55.129
On the side, right?

399
00:24:55.549 --> 00:24:58.670
And so this is the basic object in modern fluid mechanics.

400
00:24:58.830 --> 00:25:05.714
It's not the velocity of water, as though the entire ocean is a single baseball or entity that's just moving as one, right?

401
00:25:05.914 --> 00:25:14.398
No, it's different parts of the fluid are moving in different ways, and we can represent that using these little arrows that are in every single part of the fluid.

402
00:25:14.878 --> 00:25:21.516
that's actually a really important mental picture for me to kind of reframe how to think about it because my initial, um,

403
00:25:23.255 --> 00:25:49.762
mental model is this is one discrete thing and we're trying to look at it as one discrete thing yeah and what we're sort of saying is in order to get back to this idea of we want to be able to make predictions about a system uh we need to look at it from this different vantage point which is uh velocity fields which help us look at a points in space and the velocity at those points in space which then are also going to interact with each other over time exactly and

404
00:25:50.002 --> 00:25:58.513
Yeah, and so if we wanted to, for example, calculate what a dust particle is doing in that velocity field, all we have to do is follow the arrows.

405
00:25:58.573 --> 00:25:59.954
We can still recreate...

406
00:26:00.688 --> 00:26:04.909
What we want, which is like if I were to drop something, what is the position of that thing, right?

407
00:26:04.929 --> 00:26:06.289
All we have to do is follow the arrows.

408
00:26:06.609 --> 00:26:10.210
But the velocity field is something that holds all of the relevant information.

409
00:26:10.290 --> 00:26:10.890
To be able to do that.

410
00:26:10.910 --> 00:26:11.471
That's the point, yeah.

411
00:26:11.731 --> 00:26:13.791
And it becomes something that is a tenable object.

412
00:26:14.031 --> 00:26:21.013
It's something that we can understand and we don't have to worry about the 10 to the however many stuff that the fluid is made out of.

413
00:26:21.253 --> 00:26:23.953
It's less compute constrained for lack of a better way to put it.

414
00:26:23.973 --> 00:26:24.374
Yeah, yeah, yeah.

415
00:26:24.394 --> 00:26:25.074
It's one way to put it.

416
00:26:25.094 --> 00:26:26.614
Yeah, that's certainly one way to put it, yeah.

417
00:26:30.855 --> 00:26:34.036
Let's try to understand this velocity field a little bit further.

418
00:26:34.076 --> 00:26:35.357
I just showed you two examples.

419
00:26:35.397 --> 00:26:40.299
The first one was water is moving from left to right, so all of the arrows are pointing from left to right.

420
00:26:40.899 --> 00:26:49.162
Or the tornado, where the clouds are moving in a circle above you, and so all of the arrows are pointing in kind of a circle, right?

421
00:26:49.182 --> 00:26:50.103
They're like circulating.

422
00:26:51.802 --> 00:26:55.304
In both of those, we saw the particle picture, and we were like, okay, that seems reasonable.

423
00:26:55.844 --> 00:26:57.965
Now, I can imagine a fluid doing that, right?

424
00:26:58.025 --> 00:27:00.265
I can imagine the tornado, and I can imagine stuff like this.

425
00:27:00.305 --> 00:27:02.706
Now, are there velocity fields?

426
00:27:03.307 --> 00:27:09.669
Are there setups of arrows that we can make up that don't make any sense in the real world?

427
00:27:10.229 --> 00:27:10.449
Okay?

428
00:27:10.609 --> 00:27:11.369
The answer is yes.

429
00:27:11.449 --> 00:27:12.390
Let's look at these cases.

430
00:27:13.030 --> 00:27:13.830
Here are two cases.

431
00:27:14.391 --> 00:27:15.431
The first one on the left...

432
00:27:16.549 --> 00:27:20.210
is a bunch of particles that are just like emanating from the center.

433
00:27:20.430 --> 00:27:21.871
The expansion of the universe.

434
00:27:21.891 --> 00:27:23.751
Yeah, it kind of looks like the expansion of the universe.

435
00:27:23.791 --> 00:27:24.491
Exactly, right?

436
00:27:24.911 --> 00:27:30.473
It's like all of the fluid is coming from somewhere in the center, and then it's just like expanding out.

437
00:27:31.033 --> 00:27:31.273
Okay?

438
00:27:31.873 --> 00:27:32.894
How is that possible?

439
00:27:33.474 --> 00:27:34.854
Where is the fluid coming from?

440
00:27:34.914 --> 00:27:35.274
Yeah.

441
00:27:35.775 --> 00:27:35.995
Right?

442
00:27:36.275 --> 00:27:37.095
That's not possible.

443
00:27:38.387 --> 00:27:41.688
Just like in empty space, it's just like there's like just stuff.

444
00:27:41.968 --> 00:27:42.248
Right.

445
00:27:42.568 --> 00:27:43.508
Like what is the source?

446
00:27:43.548 --> 00:27:43.808
Right.

447
00:27:44.048 --> 00:27:50.570
Like if there was a hose, if I saw a bunch of particles come from the bottom and then there was like a sprinkler head or something and then it moved.

448
00:27:50.690 --> 00:27:51.430
Then I'd be like, okay.

449
00:27:51.630 --> 00:27:52.390
But here I don't see that.

450
00:27:52.470 --> 00:27:54.191
It's just the things are just coming out.

451
00:27:54.311 --> 00:27:54.931
I understood.

452
00:27:54.951 --> 00:27:55.111
Right.

453
00:27:55.191 --> 00:27:55.811
Doesn't make sense.

454
00:27:56.051 --> 00:27:56.331
Okay.

455
00:27:56.371 --> 00:27:56.611
Okay.

456
00:27:56.631 --> 00:28:01.112
First thing doesn't make sense because somehow the fluid is being created in the center and it's just moving out.

457
00:28:01.352 --> 00:28:01.533
Right.

458
00:28:01.653 --> 00:28:01.873
Okay.

459
00:28:02.433 --> 00:28:03.393
Second one on the right.

460
00:28:04.914 --> 00:28:10.337
All of the fluid is moving into a single sheet, that yellow sheet.

461
00:28:10.437 --> 00:28:11.518
It's like a sheet of paper.

462
00:28:11.938 --> 00:28:14.379
And the fluid is moving from the top into the sheet of paper.

463
00:28:14.699 --> 00:28:16.980
And the fluid is moving from the bottom into the sheet of paper.

464
00:28:17.281 --> 00:28:18.181
How is that possible?

465
00:28:18.521 --> 00:28:19.642
Where is the fluid going?

466
00:28:19.682 --> 00:28:20.102
Yeah, right.

467
00:28:20.362 --> 00:28:20.582
Right?

468
00:28:20.782 --> 00:28:20.983
Right.

469
00:28:21.223 --> 00:28:24.164
Because you would imagine it should continue.

470
00:28:24.204 --> 00:28:29.327
I mean, if I saw it shoot out to the side, then I'd be like, okay, that's possible.

471
00:28:29.347 --> 00:28:29.847
But I'm not seeing that.

472
00:28:29.867 --> 00:28:31.088
It's hitting the surface of a wall or something.

473
00:28:31.128 --> 00:28:31.228
Yeah.

474
00:28:31.388 --> 00:28:35.872
In that vector field, all of the arrows are pointing down, and there's no arrows pointing out.

475
00:28:36.292 --> 00:28:44.440
In the vector field on the left, all of the arrows are pointing out, and there's nothing that's supplying the fluid to have those arrows point out.

476
00:28:44.880 --> 00:28:51.746
And so just to be clear, we're sort of using these as sort of like an abstract theoretical or mental model to test the...

477
00:28:52.967 --> 00:28:55.249
Four corners of the sandbox of how to do this.

478
00:28:55.409 --> 00:28:56.510
Yeah.

479
00:28:56.550 --> 00:29:00.152
We're trying to understand, like, we've created this concept of a velocity field.

480
00:29:00.192 --> 00:29:00.433
Yes.

481
00:29:00.673 --> 00:29:04.475
Now, is any arbitrary vector field okay?

482
00:29:05.056 --> 00:29:05.256
Right.

483
00:29:05.556 --> 00:29:06.437
And the answer is no.

484
00:29:06.777 --> 00:29:09.759
Because on the left-hand side, that's an arbitrary vector field.

485
00:29:09.919 --> 00:29:16.944
Like, if I were to create those arrows that are just emanating from the center, if that was an electric field around a positive charge, totally fine.

486
00:29:17.185 --> 00:29:17.325
Mm-hmm.

487
00:29:17.885 --> 00:29:19.106
But for fluids, that's not fine.

488
00:29:19.166 --> 00:29:20.667
It doesn't work for fluids.

489
00:29:21.167 --> 00:29:23.749
And so we can formalize this with mathematics.

490
00:29:24.069 --> 00:29:28.773
The property that encodes this is something called the divergence of an electric field.

491
00:29:29.113 --> 00:29:29.353
Okay.

492
00:29:29.773 --> 00:29:35.438
On the left-hand side, we've got something that has a divergence of negative, negative divergence.

493
00:29:35.758 --> 00:29:38.340
That means stuff is coming in into a singular point.

494
00:29:38.400 --> 00:29:38.540
Mm-hmm.

495
00:29:39.100 --> 00:29:41.602
On the middle, we've got a positive divergence.

496
00:29:41.662 --> 00:29:43.202
I think like the collapsing of a black hole.

497
00:29:43.663 --> 00:29:44.063
Exactly.

498
00:29:44.083 --> 00:29:45.404
Just for my own visual.

499
00:29:45.424 --> 00:29:46.124
Yes, 100%.

500
00:29:46.244 --> 00:29:48.265
On the middle, that's a supernova type.

501
00:29:48.345 --> 00:29:50.687
Like the positive divergence stuff is moving out.

502
00:29:51.167 --> 00:29:53.188
And on the right, we've got a zero divergence.

503
00:29:53.749 --> 00:29:54.049
Okay.

504
00:29:54.189 --> 00:29:54.429
Okay?

505
00:29:54.889 --> 00:30:00.572
The only types of vector fields that are allowed for fluids are the ones with zero divergence.

506
00:30:01.153 --> 00:30:01.373
Okay?

507
00:30:01.613 --> 00:30:04.415
Because that means there is neither a sink nor a source.

508
00:30:04.775 --> 00:30:06.976
You can't create fluids out of nowhere.

509
00:30:07.597 --> 00:30:07.837
So...

510
00:30:08.637 --> 00:30:12.000
Whatever fluid you have, it's got to come from somewhere and it's got to be going somewhere.

511
00:30:12.701 --> 00:30:12.921
Okay?

512
00:30:13.381 --> 00:30:15.023
So this is a key constraint.

513
00:30:17.191 --> 00:30:18.431
For our vector field.

514
00:30:18.751 --> 00:30:20.832
For fluids specifically.

515
00:30:20.852 --> 00:30:21.772
For fluids specifically.

516
00:30:21.792 --> 00:30:23.052
For other things, it's totally fine.

517
00:30:23.072 --> 00:30:27.494
Like, for example, if we were talking about the electric field, on the left would be a negative charge.

518
00:30:27.814 --> 00:30:29.254
In the middle would be a positive charge.

519
00:30:29.414 --> 00:30:29.654
Fine.

520
00:30:29.774 --> 00:30:30.034
Fine.

521
00:30:30.314 --> 00:30:30.854
Totally fine.

522
00:30:31.335 --> 00:30:33.855
But for fluids, that can't be possible.

523
00:30:33.995 --> 00:30:35.976
So I think this is an important point here.

524
00:30:35.996 --> 00:30:42.857
The idea is this concept of divergence applies differently based on the underlying subject matter that you're speaking to.

525
00:30:43.097 --> 00:30:44.778
And so as we've gone through this progression over...

526
00:30:45.678 --> 00:30:52.084
of testing vector fields, it only can work when the divergence is zero.

527
00:30:52.144 --> 00:30:52.805
Is zero, yeah.

528
00:30:53.185 --> 00:30:55.287
Any non-zero divergence and it doesn't make sense.

529
00:30:55.607 --> 00:30:58.870
Another way of putting this is something called incompressibility.

530
00:30:59.190 --> 00:31:01.032
You cannot compress the fluid.

531
00:31:01.052 --> 00:31:01.252
Okay.

532
00:31:01.392 --> 00:31:01.612
Okay.

533
00:31:01.893 --> 00:31:03.114
Those are the fluids that we're talking about.

534
00:31:03.134 --> 00:31:03.955
And if we go to the next...

535
00:31:04.515 --> 00:31:04.836
Slide.

536
00:31:05.056 --> 00:31:06.698
Those are our two Navier-Stokes equations.

537
00:31:06.738 --> 00:31:08.461
You can already see that we've already got one of them.

538
00:31:08.741 --> 00:31:09.042
Okay.

539
00:31:09.262 --> 00:31:09.522
Okay.

540
00:31:09.703 --> 00:31:14.570
The second one is that the divergence of this fluid velocity field is zero.

541
00:31:14.730 --> 00:31:15.331
Yeah.

542
00:31:15.351 --> 00:31:17.354
That's already the second Navier-Stokes equation.

543
00:31:17.394 --> 00:31:17.654
Okay.

544
00:31:17.874 --> 00:31:19.697
So we're already, you would say, halfway through.

545
00:31:19.717 --> 00:31:19.817
Okay.

546
00:31:20.698 --> 00:31:26.641
And can you remind me again, we have our, I guess, delta symbol and our U symbol equals zero.

547
00:31:26.661 --> 00:31:27.682
That's what you're referring to.

548
00:31:27.722 --> 00:31:29.002
Yeah, that's the second one.

549
00:31:29.363 --> 00:31:30.263
The second one there.

550
00:31:30.443 --> 00:31:33.085
You would say the divergence of U is zero.

551
00:31:33.305 --> 00:31:37.007
Right, which was that diagram where we saw the center point, everything moving in one direction.

552
00:31:37.027 --> 00:31:38.067
Everything moving in one direction.

553
00:31:38.347 --> 00:31:41.789
Stuff came in, stuff is going out, and everything is fine.

554
00:31:41.809 --> 00:31:47.152
You're not creating or destroying fluid out of...

555
00:31:47.752 --> 00:31:48.852
I don't know where.

556
00:31:48.972 --> 00:31:49.172
Yeah.

557
00:31:49.312 --> 00:31:54.574
And for engineers, you might you might understand this with the continuity equation.

558
00:31:54.914 --> 00:32:05.017
Like, for example, this is applied very directly if you've got a pipe that has a large cross section and you've got water coming in and then the pipe like narrows into a small cross section.

559
00:32:05.197 --> 00:32:08.038
For our audio listeners, imagine a wine bottle on its side.

560
00:32:08.138 --> 00:32:08.838
Yes, exactly.

561
00:32:08.878 --> 00:32:10.919
Wine bottle on the side, but its base has been cut off.

562
00:32:11.019 --> 00:32:11.259
Yes.

563
00:32:11.339 --> 00:32:14.961
And then now you've got water or wine coming in from the base, right?

564
00:32:15.341 --> 00:32:22.724
As it exits the nozzle or as it exits the tip, it's going to be moving faster because all of that fluid has been jammed in.

565
00:32:23.524 --> 00:32:25.365
You cannot create or destroy fluid.

566
00:32:25.665 --> 00:32:29.286
So the amount of fluid that's coming in has to equal the amount of fluid going out.

567
00:32:29.326 --> 00:32:35.148
But that means that if I've got a larger cross-section, then in a single amount of time, I've got some amount of volume coming in.

568
00:32:35.488 --> 00:32:40.570
And with a smaller cross-section, that same amount of volume has to go out, which means the velocity has to increase.

569
00:32:40.810 --> 00:32:43.611
This is the continuity equation in action.

570
00:32:44.131 --> 00:32:48.893
And a mathematically succinct way of putting it is that del.u equals zero.

571
00:32:48.933 --> 00:32:51.274
The divergence of the vector field is always zero.

572
00:32:52.114 --> 00:33:01.677
And it kind of makes intuitive sense for someone like me, which is when I look at the sort of the area of the cross section in the wider part, there's just more space for stuff to move.

573
00:33:02.017 --> 00:33:04.457
And we have to move the same amount of volume through a smaller space.

574
00:33:04.537 --> 00:33:04.777
Yeah.

575
00:33:04.857 --> 00:33:12.819
So it necessarily needs to accelerate right at that point in order to for the divergence of you to equal zero.

576
00:33:12.839 --> 00:33:13.080
Yeah.

577
00:33:13.120 --> 00:33:13.800
For that fact.

578
00:33:15.080 --> 00:33:18.701
To be true, you have to account for this difference in velocity.

579
00:33:18.901 --> 00:33:19.301
Right.

580
00:33:19.341 --> 00:33:27.042
Now, crucially, this is an independent thing from that first Navier-Stokes equation, which is going to be the topic for the next segment.

581
00:33:27.182 --> 00:33:27.843
Okay.

582
00:33:27.903 --> 00:33:29.463
This is an independent criteria.

583
00:33:29.583 --> 00:33:29.863
Okay.

584
00:33:29.943 --> 00:33:34.164
If we didn't have this, you could have different types of fluids.

585
00:33:34.904 --> 00:33:37.044
You could have, for example, compressible fluids.

586
00:33:37.664 --> 00:33:40.125
Like air, for example, is kind of compressible.

587
00:33:40.818 --> 00:33:45.744
In the sense that if you heat it up, it's going to expand, and if you cool it down, it's going to contract.

588
00:33:46.184 --> 00:33:49.027
Now, that's not a divergence-free fluid.

589
00:33:49.388 --> 00:33:56.215
The Navier-Stokes equations, these two, have to do with fluids in the sense of water that is really not compressible.

590
00:33:56.255 --> 00:33:57.657
The density is the same everywhere.

591
00:33:57.937 --> 00:33:59.299
You can't adjust the density.

592
00:33:59.639 --> 00:34:05.701
So a key takeaway for – one of the key takeaways for this is the concept of this idea of incompressible.

593
00:34:05.801 --> 00:34:06.021
Yes.

594
00:34:06.081 --> 00:34:09.963
Which we've just defined from mathematical first principles.

595
00:34:10.003 --> 00:34:10.563
Yeah.

596
00:34:10.623 --> 00:34:14.044
But it really just means you can't – you know, you can't pack stuff in.

597
00:34:14.424 --> 00:34:14.544
Right.

598
00:34:14.564 --> 00:34:15.945
The density is always constant.

599
00:34:16.185 --> 00:34:16.345
Right.

600
00:34:16.405 --> 00:34:16.545
Right?

601
00:34:16.765 --> 00:34:18.846
It's like one of those – have you seen those like stress thingies?

602
00:34:18.866 --> 00:34:19.766
Yeah, the little stress ball.

603
00:34:19.786 --> 00:34:23.727
Yeah, where you like squeeze it and then the thing is going to like shoot out somewhere else.

604
00:34:24.127 --> 00:34:25.528
That's an incompressible fluid.

605
00:34:25.888 --> 00:34:26.068
Right?

606
00:34:26.108 --> 00:34:28.469
Because if I squish it, it's got to go somewhere else.

607
00:34:28.549 --> 00:34:28.769
Mm-hmm.

608
00:34:29.289 --> 00:34:29.470
Right.

609
00:34:29.650 --> 00:34:29.890
Okay.

610
00:34:29.950 --> 00:34:32.152
It's not like one of those foam things where you can just compress it.

611
00:34:33.634 --> 00:34:35.155
Right, which is actually an important distinction.

612
00:34:35.175 --> 00:34:42.743
Okay, so this is helpful because that's one half of what people say when they say the Navier-Stokes equations.

613
00:34:43.083 --> 00:34:48.849
And it's going to be the basis by which I'm guessing we're now going to build an understanding of the second equation.

614
00:34:48.969 --> 00:34:49.949
Of the second one, right?

615
00:34:50.409 --> 00:34:57.912
We've sort of gotten an intuitive understanding of what types of velocity fields are allowed, first of all.

616
00:34:58.092 --> 00:34:58.432
Right.

617
00:34:58.592 --> 00:35:04.294
And kind of what the velocity field represents, which is a bunch of these arrows that tells you how the velocity…

618
00:35:05.185 --> 00:35:07.067
is at that moment in space and time.

619
00:35:07.287 --> 00:35:07.827
In a fluid.

620
00:35:07.887 --> 00:35:08.708
In a fluid, right?

621
00:35:08.768 --> 00:35:09.929
In an incompressible fluid.

622
00:35:10.189 --> 00:35:11.270
In an incompressible fluid.

623
00:35:11.290 --> 00:35:11.470
Yeah.

624
00:35:11.910 --> 00:35:15.713
Okay, so now let's try to apply this velocity field to Newton's laws.

625
00:35:16.053 --> 00:35:16.314
Okay.

626
00:35:16.574 --> 00:35:17.555
Or the other way, actually.

627
00:35:17.875 --> 00:35:20.557
We want to apply Newton's laws to this velocity field.

628
00:35:20.877 --> 00:35:22.098
Now, that's easier said than done.

629
00:35:23.019 --> 00:35:28.603
The challenge now is how do we talk about Newton's laws, which has to do with individual particles in the fluid, right?

630
00:35:28.943 --> 00:35:33.507
It's like this water molecule is getting forces from other water molecules, so it's moving around.

631
00:35:34.380 --> 00:35:36.100
Newton's laws has to do with the stuff.

632
00:35:36.901 --> 00:35:38.801
It doesn't have to do with the velocity field.

633
00:35:39.201 --> 00:35:44.102
So we have to now put Newton's laws in the language that we've just invented.

634
00:35:44.843 --> 00:35:55.865
And just to take a quick step back, just from my brain, part of, like, it's like, you know, if you have a bathtub with a little boat, toy boat in it for a kid, right?

635
00:35:57.046 --> 00:36:01.147
We talked about Newton's laws in the context of, like, the boat in the water.

636
00:36:01.227 --> 00:36:01.367
Yeah.

637
00:36:01.387 --> 00:36:01.487
Yeah.

638
00:36:02.214 --> 00:36:05.357
And then when we talked about velocity fields, it's like the medium of the water.

639
00:36:05.397 --> 00:36:06.438
Yeah, it's the water itself.

640
00:36:06.458 --> 00:36:07.258
It's the water itself.

641
00:36:07.338 --> 00:36:08.379
And these are different.

642
00:36:08.399 --> 00:36:14.004
So it's like the discrete object versus the medium in which an object or many objects may exist in.

643
00:36:14.024 --> 00:36:14.184
Yeah.

644
00:36:14.584 --> 00:36:18.588
And the lens in which we look at them is from different doorways.

645
00:36:18.648 --> 00:36:18.908
Yes.

646
00:36:19.529 --> 00:36:23.092
In the first one, I could easily just be like, what is the position of the boat?

647
00:36:23.332 --> 00:36:23.452
Right.

648
00:36:23.472 --> 00:36:24.193
Oh, it's like right there.

649
00:36:24.213 --> 00:36:24.493
Right.

650
00:36:24.513 --> 00:36:24.633
Right.

651
00:36:24.853 --> 00:36:26.134
What is the position of the water?

652
00:36:26.414 --> 00:36:27.054
What do you mean?

653
00:36:27.354 --> 00:36:27.974
Which water?

654
00:36:28.415 --> 00:36:29.075
Yes.

655
00:36:29.835 --> 00:36:33.197
There's an Avogadro's number of water molecules that we're talking about.

656
00:36:33.477 --> 00:36:33.597
Right?

657
00:36:33.637 --> 00:36:36.638
So instead we have to reinvent this new language of the velocity field.

658
00:36:36.998 --> 00:36:39.159
And now what we're going to do is build this step by step.

659
00:36:39.419 --> 00:36:42.901
We're going to apply Newton's second law, which is force equals mass times acceleration.

660
00:36:43.241 --> 00:36:44.942
We're going to apply that to our velocity field.

661
00:36:44.962 --> 00:36:52.165
The first thing we're going to do is try and describe what is the acceleration of a fluid particle.

662
00:36:53.098 --> 00:36:53.438
Okay.

663
00:36:53.658 --> 00:36:55.059
Like in this velocity field.

664
00:36:55.299 --> 00:36:55.599
Okay.

665
00:36:55.719 --> 00:36:55.979
Okay.

666
00:36:56.139 --> 00:36:56.379
Okay.

667
00:36:56.499 --> 00:37:04.401
And we're going to use traffic as an analogy.

668
00:37:04.581 --> 00:37:08.062
If you live in LA, you very much are intuitively familiar.

669
00:37:08.162 --> 00:37:08.402
Yes.

670
00:37:08.502 --> 00:37:12.103
And you're probably listening to this while you're in traffic.

671
00:37:12.623 --> 00:37:12.903
Okay.

672
00:37:13.283 --> 00:37:19.665
So traffic is kind of a nice analogy for fluids because you've got these working agents and

673
00:37:20.220 --> 00:37:21.141
in traffic, i.e.

674
00:37:21.161 --> 00:37:22.281
the cars, right?

675
00:37:22.301 --> 00:37:31.366
The cars can be the individual fluid particles, and the highway is like the receptacle, and the entire flow of traffic is the fluid, okay?

676
00:37:32.726 --> 00:37:38.309
The challenge now is that we want to have a description for how the individual cars accelerate,

677
00:37:38.911 --> 00:37:45.233
or de-accelerate if they brake or if they press on the pedal, based on the stuff that we can see from a news helicopter.

678
00:37:45.713 --> 00:37:51.334
Because the news helicopter is up there, and it's seeing the velocity field of these cars, right?

679
00:37:51.394 --> 00:37:54.175
It's seeing, like, oh, over there, cars are moving fast.

680
00:37:54.335 --> 00:37:56.556
Over here, the cars are moving slower.

681
00:37:56.656 --> 00:38:03.638
And, like, that's how they, like, if you're on the radio, they tell you, oh, the 101 is backed up from Sepulveda, blah, blah, blah, blah, blah, right?

682
00:38:04.278 --> 00:38:04.498
Um...

683
00:38:05.573 --> 00:38:09.579
So they're actually watching the cars, but they're watching an aggregate mass of cars.

684
00:38:09.619 --> 00:38:14.365
It's not like they're following a single car and seeing how stuff is happening.

685
00:38:14.786 --> 00:38:20.293
So what we want to do is describe the acceleration of a single car based on what we see.

686
00:38:21.264 --> 00:38:22.685
from a chopper, right?

687
00:38:22.845 --> 00:38:25.648
Now, first, let's think about how cars can accelerate.

688
00:38:25.688 --> 00:38:31.732
Cars can accelerate and deaccelerate as a response to the traffic itself changing at a specific spot.

689
00:38:31.792 --> 00:38:39.098
For example, suppose you're in the chopper and you notice that everyone is moving at like 80 miles an hour, right?

690
00:38:39.178 --> 00:38:43.562
It's a nice day and everyone's moving at 80 miles an hour and all of a sudden it starts raining, right?

691
00:38:44.101 --> 00:38:46.542
I was going to say 80 miles an hour in LA never happens.

692
00:38:46.562 --> 00:38:47.543
Yeah, yeah, never happens.

693
00:38:47.563 --> 00:38:48.824
This is an unrealistic example.

694
00:38:48.844 --> 00:38:50.485
Yeah, yeah, yeah.

695
00:38:50.665 --> 00:38:51.605
That's totally fair.

696
00:38:52.005 --> 00:38:53.947
Anyways, that's what I'm stuck with in the visuals.

697
00:38:54.207 --> 00:38:57.869
So we got 80 miles an hour, and then all of a sudden, it starts raining.

698
00:38:58.489 --> 00:38:58.729
Okay?

699
00:38:59.009 --> 00:39:02.391
And within 10 seconds, everyone slows down to 60 miles an hour.

700
00:39:03.092 --> 00:39:05.333
Obviously, the cars have deaccelerated, right?

701
00:39:05.373 --> 00:39:08.375
They've used their brakes to slow down.

702
00:39:09.235 --> 00:39:13.600
This 20 mile per hour reduction happened in 10 seconds, let's say.

703
00:39:14.201 --> 00:39:18.806
And so the cars accelerated, de-accelerated at negative two miles per hour per second.

704
00:39:19.187 --> 00:39:24.252
They're slowing as a response to what the traffic itself is doing.

705
00:39:24.292 --> 00:39:25.974
The entire traffic slowed down.

706
00:39:26.911 --> 00:39:28.732
And so we slowed down.

707
00:39:28.992 --> 00:39:36.314
It's kind of like seeing, like, you know, in the traffic map of LA, everything was green, and then rain started happening, everything turned yellow.

708
00:39:36.914 --> 00:39:40.075
Well, that means all of the individual cars also deaccelerated.

709
00:39:40.436 --> 00:39:42.796
So that's one way that you can deaccelerate.

710
00:39:42.836 --> 00:39:50.119
And this is called the Eulerian perspective, because it's asking how the field itself is changing as a function of time at a given spot.

711
00:39:50.659 --> 00:39:55.081
And this is represented by the derivative of the velocity field with respect to time.

712
00:39:55.601 --> 00:39:55.921
This is...

713
00:39:56.361 --> 00:39:57.162
It makes sense, right?

714
00:39:57.182 --> 00:39:58.443
It's a velocity field.

715
00:39:58.843 --> 00:40:01.446
If I take the derivative of velocity, I get acceleration.

716
00:40:01.866 --> 00:40:02.447
And there you go.

717
00:40:02.487 --> 00:40:07.231
This is an acceleration of my particles, right?

718
00:40:07.511 --> 00:40:14.898
In this case, the thing is slowing down, but you can easily imagine after the rain ends, the things go back to 80 miles per hour, and it's going to speed up again.

719
00:40:15.238 --> 00:40:18.421
This is represented by a derivative of my velocity field.

720
00:40:19.101 --> 00:40:23.045
And so the idea is, here's a way to describe the velocity field as...

721
00:40:23.245 --> 00:40:27.686
as this system unique and different from the discrete individual car in traffic.

722
00:40:27.706 --> 00:40:27.946
Yeah.

723
00:40:27.986 --> 00:40:28.206
Yeah.

724
00:40:28.306 --> 00:40:37.609
And we can use the description of the velocity field and how the entire field is changing in time to describe how an individual car is experiencing acceleration.

725
00:40:37.929 --> 00:40:38.789
That's the key, right?

726
00:40:39.330 --> 00:40:39.870
So good.

727
00:40:39.990 --> 00:40:43.751
We've got at least some of acceleration down, but that can't be the whole story.

728
00:40:44.231 --> 00:40:44.491
Okay.

729
00:40:45.031 --> 00:40:46.091
That cannot be the whole story.

730
00:40:46.452 --> 00:40:48.652
For example, let's take a look at this airfoil.

731
00:40:49.873 --> 00:40:51.373
This is a airplane wing.

732
00:40:52.802 --> 00:40:57.270
where the air is coming in from the left and it's going over the wing and under the wing and then to the right.

733
00:40:57.730 --> 00:40:57.971
Okay?

734
00:40:58.552 --> 00:41:05.103
Now, crucially, on the lower right, you're seeing the velocity field representation of this airplane wing.

735
00:41:05.123 --> 00:41:05.263
Mm-hmm.

736
00:41:05.548 --> 00:41:05.788
Okay.

737
00:41:06.048 --> 00:41:13.131
So the arrows above the wing, the velocity is very high because the air is like getting pushed up and then there's high pressure.

738
00:41:13.171 --> 00:41:15.492
So it's like, these are the red arrows.

739
00:41:15.512 --> 00:41:15.692
Yeah.

740
00:41:15.712 --> 00:41:17.152
Those are the red arrows above the wing.

741
00:41:17.172 --> 00:41:20.233
So there's higher pressure above the wing, lower pressure below the wing.

742
00:41:20.513 --> 00:41:22.734
And so there's slower fluid below the wing.

743
00:41:23.474 --> 00:41:28.036
And just the length of these arrows is meaningful here in that there's more.

744
00:41:28.236 --> 00:41:28.676
Yeah.

745
00:41:28.736 --> 00:41:29.416
It's faster.

746
00:41:29.516 --> 00:41:29.977
Faster.

747
00:41:30.137 --> 00:41:30.997
Yeah.

748
00:41:31.617 --> 00:41:32.457
And okay.

749
00:41:32.477 --> 00:41:33.578
I'm already seeing where this is going.

750
00:41:33.598 --> 00:41:33.778
Okay.

751
00:41:33.798 --> 00:41:33.898
Okay.

752
00:41:34.178 --> 00:41:37.600
So this is the, on the left is your particles that are moving.

753
00:41:38.060 --> 00:41:38.280
Right.

754
00:41:38.600 --> 00:41:39.640
Like our flow of traffic.

755
00:41:39.660 --> 00:41:40.741
Yeah, our flow of traffic.

756
00:41:41.181 --> 00:41:44.803
And on the right-hand side is the velocity field.

757
00:41:45.563 --> 00:41:53.847
But if I were to take a derivative of that velocity field, if I were to ask, how is this velocity field changing with respect to time, the answer would be zero.

758
00:41:54.167 --> 00:41:54.447
Hmm.

759
00:41:55.007 --> 00:41:57.168
Because this is a static velocity field.

760
00:41:57.188 --> 00:42:11.953
If you go back to that picture that I had with the tornado, even though the particles were circling around, the velocity field was static because in this point, the velocity is this way, and then as it goes around, it's like coming around, right?

761
00:42:12.573 --> 00:42:14.214
And so what?

762
00:42:14.354 --> 00:42:17.135
Does that mean that the particles are not moving around?

763
00:42:18.550 --> 00:42:24.794
Another way of putting it with the chopper idea is like, you know, you have, let's say, an accident over here.

764
00:42:25.714 --> 00:42:25.954
Okay?

765
00:42:26.455 --> 00:42:27.195
On the 101.

766
00:42:28.976 --> 00:42:31.458
Way behind, way ahead of the accident...

767
00:42:32.266 --> 00:42:35.388
the speed is very high.

768
00:42:35.608 --> 00:42:38.269
But as you approach the accident, it's very low, right?

769
00:42:38.549 --> 00:42:42.871
But the accident isn't going anywhere because the police are very far away.

770
00:42:42.891 --> 00:42:43.592
It just happened.

771
00:42:44.012 --> 00:42:45.332
And so the police haven't gotten there.

772
00:42:45.573 --> 00:42:50.955
So the accident is not going anywhere, which means that ahead of the accident, the speed is very high.

773
00:42:51.415 --> 00:42:52.976
Before the accident, the speed is very low.

774
00:42:53.056 --> 00:42:54.197
And that's not changing.

775
00:42:54.777 --> 00:42:57.079
The traffic pattern is not changing.

776
00:42:57.220 --> 00:42:59.182
It'll still be red here, green here.

777
00:42:59.462 --> 00:43:05.028
If I come back five minutes later, because of how slow the response is, it'll still be red here and green here.

778
00:43:05.428 --> 00:43:08.672
So the traffic pattern has not changed in time.

779
00:43:08.932 --> 00:43:09.412
And yet...

780
00:43:10.407 --> 00:43:15.711
I think you would be remiss to say, yeah, the cars are accelerating.

781
00:43:16.072 --> 00:43:21.396
Once you get past the accident and you're done rubbernecking, yeah, you're like, okay, I got to get on with my life.

782
00:43:21.456 --> 00:43:22.477
Like, let's go, right?

783
00:43:22.837 --> 00:43:23.077
Okay.

784
00:43:23.558 --> 00:43:28.421
And so I just want to make sure that I'm following this here because part of what we're trying to say is –

785
00:43:29.302 --> 00:43:37.708
We're trying to go from the velocity field and then say, can we describe an individual object within that velocity field?

786
00:43:37.728 --> 00:43:38.548
And how it accelerates.

787
00:43:38.568 --> 00:43:41.911
And how it accelerates based on the mapping we see in the velocity field.

788
00:43:41.931 --> 00:43:42.611
Yes, exactly.

789
00:43:42.671 --> 00:43:52.018
And part of what you're pointing out here is there are contexts like the airfoil or the traffic accident example you just described where the velocity field…

790
00:43:52.718 --> 00:43:53.459
equals zero.

791
00:43:54.139 --> 00:43:55.520
And so that means you can't...

792
00:43:55.821 --> 00:43:56.701
The derivative of...

793
00:43:56.721 --> 00:43:58.503
The velocity field is not changing.

794
00:43:59.123 --> 00:43:59.544
Sorry, right.

795
00:43:59.604 --> 00:44:01.746
So the velocity field, it's static.

796
00:44:01.826 --> 00:44:02.066
Yeah.

797
00:44:02.346 --> 00:44:04.328
It's not changing, which means you can...

798
00:44:05.298 --> 00:44:11.104
In some cases, you may not be able to accurately describe certain particles within that context.

799
00:44:11.144 --> 00:44:14.187
Yeah, using just the time derivative of the velocity field.

800
00:44:14.207 --> 00:44:17.930
You have to figure out something else about the velocity field that's doing this.

801
00:44:17.950 --> 00:44:18.931
And let's just think about this.

802
00:44:18.971 --> 00:44:22.334
I think we've got in the next one, you know, the highway slowdown.

803
00:44:22.494 --> 00:44:26.178
Let's say there's a – let's say there's – Next one here.

804
00:44:26.238 --> 00:44:27.319
And the next one, exactly.

805
00:44:27.579 --> 00:44:29.481
So let's say there's road work ahead, right?

806
00:44:29.501 --> 00:44:29.621
Okay.

807
00:44:30.883 --> 00:44:32.204
A car is going to slow down.

808
00:44:33.045 --> 00:44:38.290
Now, all of the cars near the roadwork are going to be going slow, and the cars before the roadwork are going to be going fast.

809
00:44:38.691 --> 00:44:40.292
The roadwork doesn't change where it is.

810
00:44:40.873 --> 00:44:42.695
But what is making the car slow down?

811
00:44:42.795 --> 00:44:45.558
It's the fact that it is approaching the roadwork.

812
00:44:46.099 --> 00:44:48.421
There is a spatial difference between...

813
00:44:49.615 --> 00:44:50.596
In the velocity field.

814
00:44:51.136 --> 00:44:51.957
You see?

815
00:44:51.997 --> 00:44:56.680
There's the time difference that we talked about earlier, where the rain came in, so everyone slowed down.

816
00:44:57.041 --> 00:45:02.565
People slowed down here if there's only a patch of rain here, and everyone else is fine, but this patch of rain, people slowed down.

817
00:45:03.005 --> 00:45:04.686
So that's a single spot.

818
00:45:05.467 --> 00:45:06.268
Stuff is changing.

819
00:45:06.688 --> 00:45:13.973
But here, I am accelerating or deaccelerating because I'm moving through the traffic pattern.

820
00:45:14.694 --> 00:45:18.477
If I go from a red to a green, I'm going to accelerate in my map.

821
00:45:19.361 --> 00:45:26.808
If I go from a green to a red, I'm going to de-accelerate, even though where the green and red are, are staying exactly the same.

822
00:45:26.828 --> 00:45:30.772
The fact that I'm moving through it is making me accelerate or de-accelerate.

823
00:45:31.448 --> 00:45:39.990
And so I think part of what you're trying to get at here is that the time derivative of the velocity field is not sufficient to describe the acceleration of an individual object.

824
00:45:40.210 --> 00:45:41.610
Yeah, it's part of the story.

825
00:45:41.850 --> 00:45:50.332
But it seems like the distance is potentially the other ingredient in the pie, at least based on the example that you just brought up.

826
00:45:50.492 --> 00:45:51.153
Exactly, yeah.

827
00:45:51.233 --> 00:45:52.133
And that's a big part of it.

828
00:45:52.173 --> 00:45:54.153
It's the spatial difference.

829
00:45:54.353 --> 00:45:57.694
It's how the velocity field changes as I move from one space to another.

830
00:46:01.755 --> 00:46:05.238
You might think, well, what if I took a time derivative before?

831
00:46:05.658 --> 00:46:08.381
What if I take a spatial derivative of the velocity field?

832
00:46:08.401 --> 00:46:12.244
So that tells me how much the velocity is changing from one spot to another, right?

833
00:46:12.564 --> 00:46:19.410
If I move here and then I move like 100 feet over there, if I've slowed down, that tells me how much I should slow down.

834
00:46:19.750 --> 00:46:20.570
That's not everything.

835
00:46:20.811 --> 00:46:21.051
Okay.

836
00:46:21.211 --> 00:46:21.451
Okay?

837
00:46:21.871 --> 00:46:23.393
There is a secondary effect.

838
00:46:23.493 --> 00:46:23.793
Okay.

839
00:46:24.333 --> 00:46:28.577
And to understand that, let's look at that scenario again where you've got road work ahead.

840
00:46:28.597 --> 00:46:28.817
Okay.

841
00:46:28.837 --> 00:46:29.037
Okay?

842
00:46:29.197 --> 00:46:29.418
Okay.

843
00:46:30.158 --> 00:46:32.842
But now I've got two different cars that are approaching.

844
00:46:32.902 --> 00:46:36.847
The first car approaches that road work at 100 meters per second.

845
00:46:36.867 --> 00:46:37.428
Okay.

846
00:46:37.808 --> 00:46:38.068
Okay?

847
00:46:39.350 --> 00:46:43.876
And the second car approaches that road work at exactly half the speed, 50 meters per second.

848
00:46:44.557 --> 00:46:44.677
Hmm.

849
00:46:46.097 --> 00:46:50.418
I think you'll agree that the guy that is moving faster is going to have to break harder.

850
00:46:50.438 --> 00:46:50.698
Yeah.

851
00:46:51.098 --> 00:46:51.378
Yeah.

852
00:46:51.578 --> 00:46:57.020
Because he's going to come upon that road work way faster than the guy who's slowing.

853
00:46:57.100 --> 00:47:00.441
He's going to have to be like, oh, there's road work.

854
00:47:00.801 --> 00:47:02.061
Fines are doubled.

855
00:47:02.101 --> 00:47:03.841
I better slow down a lot.

856
00:47:03.961 --> 00:47:05.282
So there's a secondary effect.

857
00:47:05.722 --> 00:47:09.883
It has to do not just with how fast the traffic pattern is changing.

858
00:47:11.355 --> 00:47:14.756
but also how fast the individual car is moving.

859
00:47:17.037 --> 00:47:17.257
Yes?

860
00:47:17.457 --> 00:47:17.677
Yes.

861
00:47:17.917 --> 00:47:22.859
Because, so, and again, trying to use the terminology again, we talked about the time derivative matters.

862
00:47:23.279 --> 00:47:27.320
We talked about the spatial matters.

863
00:47:27.721 --> 00:47:31.562
And also the individual object's velocity.

864
00:47:31.662 --> 00:47:32.682
Velocity.

865
00:47:32.802 --> 00:47:33.223
The speed.

866
00:47:33.263 --> 00:47:34.103
The speed itself.

867
00:47:34.763 --> 00:47:39.844
is the third item in our cookie recipe here.

868
00:47:40.044 --> 00:47:50.447
Like all of these three things are relevant in order to be able to go from looking at a velocity field and describing an individual object within that velocity field.

869
00:47:50.527 --> 00:47:51.267
Acceleration.

870
00:47:51.287 --> 00:47:52.307
Acceleration, exactly.

871
00:47:52.647 --> 00:48:04.270
And so for the second scenario, we have to take into account both the velocity and we have to take into account the change in velocity.

872
00:48:05.095 --> 00:48:05.315
Right.

873
00:48:05.496 --> 00:48:05.776
Okay?

874
00:48:05.876 --> 00:48:06.016
Right.

875
00:48:06.036 --> 00:48:10.783
The change in velocity as a function of distance is like how far away is the road work?

876
00:48:11.083 --> 00:48:11.244
Right.

877
00:48:11.484 --> 00:48:16.411
And my own velocity as I approach that also has something to do with how fast I need to break.

878
00:48:16.471 --> 00:48:16.651
Yes.

879
00:48:16.851 --> 00:48:21.478
So it's going to be a multiplication of the velocity field times the spatial derivative.

880
00:48:21.498 --> 00:48:21.638
Yeah.

881
00:48:22.051 --> 00:48:25.453
That's the upside-down triangle.

882
00:48:25.873 --> 00:48:26.714
It's called Nambla.

883
00:48:26.854 --> 00:48:27.514
I call it del.

884
00:48:27.754 --> 00:48:29.755
But it's u dot del times u.

885
00:48:30.416 --> 00:48:30.676
Okay?

886
00:48:31.396 --> 00:48:32.817
This is called the convective term.

887
00:48:33.598 --> 00:48:43.343
And what this convective term is doing now, it's giving us in the example we just talked about here, it's giving us the 100 meters per second, 80 meters per second.

888
00:48:43.743 --> 00:48:43.903
Right.

889
00:48:43.983 --> 00:48:44.364
Okay.

890
00:48:44.584 --> 00:48:48.226
Multiplied by the del u, which is the spatial delu,

891
00:48:50.348 --> 00:48:52.170
How much time do I have before the road work?

892
00:48:52.270 --> 00:48:55.553
I mean, how much distance do I have before the road work gets there, right?

893
00:48:55.754 --> 00:48:58.236
How much time do I have is dependent on how fast I was getting there.

894
00:48:58.256 --> 00:48:58.877
Yeah, yeah, yeah.

895
00:48:58.977 --> 00:48:59.257
Okay.

896
00:48:59.638 --> 00:48:59.938
Okay.

897
00:49:00.078 --> 00:49:00.979
And that's the okay.

898
00:49:00.999 --> 00:49:01.359
Make sense?

899
00:49:01.480 --> 00:49:01.700
Yes.

900
00:49:02.020 --> 00:49:02.340
Okay.

901
00:49:02.421 --> 00:49:04.182
And both of these terms combined...

902
00:49:04.963 --> 00:49:06.625
give us our acceleration.

903
00:49:07.546 --> 00:49:11.449
We've got the Eulerian perspective, which is how is the traffic pattern itself changing?

904
00:49:11.830 --> 00:49:19.036
For example, is it going from green to red if rain came down or something, or if there was an accident, right?

905
00:49:19.317 --> 00:49:24.862
Then at a specific spot, that highway would go from green to red on our Google Maps.

906
00:49:24.982 --> 00:49:26.323
And that's the first term.

907
00:49:26.403 --> 00:49:29.086
That's our velocity, the derivative of the velocity field.

908
00:49:29.386 --> 00:49:30.608
Yeah, with respect to time, right?

909
00:49:30.628 --> 00:49:31.229
With respect to time.

910
00:49:31.249 --> 00:49:33.071
Because the field itself is changing.

911
00:49:33.091 --> 00:49:33.732
The time derivative.

912
00:49:33.752 --> 00:49:35.815
The color is changing on the Google Maps.

913
00:49:35.875 --> 00:49:36.055
Right.

914
00:49:36.315 --> 00:49:36.495
Right?

915
00:49:36.736 --> 00:49:40.821
On the right-hand side, the Google Maps color is staying the same, but I am moving through.

916
00:49:41.642 --> 00:49:44.165
That portion of the yellow or the red.

917
00:49:44.205 --> 00:49:44.345
Yeah.

918
00:49:44.366 --> 00:49:44.466
Yeah.

919
00:49:44.906 --> 00:49:47.908
And so I'm going to be breaking or accelerating based on that.

920
00:49:47.968 --> 00:49:49.449
And effectively a more granular way.

921
00:49:49.469 --> 00:49:53.292
There's like a bird's eye and then there's a zoom.

922
00:49:53.412 --> 00:49:53.632
Yes.

923
00:49:53.812 --> 00:49:54.793
And both of those matter.

924
00:49:54.833 --> 00:49:58.716
And that's why it's called actually the Eulerian perspective because that's kind of the bird's eye view.

925
00:49:58.996 --> 00:50:02.518
And then the Lagrangian perspective is like the nitty gritty.

926
00:50:02.598 --> 00:50:03.359
What's happening there?

927
00:50:03.679 --> 00:50:05.040
And that's the conceptual trick.

928
00:50:05.180 --> 00:50:05.460
Okay.

929
00:50:05.560 --> 00:50:05.821
Okay.

930
00:50:05.841 --> 00:50:05.981
And.

931
00:50:07.302 --> 00:50:14.730
In the Navier-Stokes equations, we're writing Newton's laws for the person in the car, for the car itself, based on the data that's collected by the chopper.

932
00:50:15.511 --> 00:50:18.134
The chopper is only finding you.

933
00:50:18.154 --> 00:50:22.599
The chopper only knows the green and the red and everything like that.

934
00:50:22.659 --> 00:50:23.921
It's seeing these aggregate...

935
00:50:24.813 --> 00:50:25.594
traffic patterns.

936
00:50:25.994 --> 00:50:33.819
But from those aggregate traffic patterns, I could tell you if a car is over there, how fast is it going to have to accelerate or deaccelerate?

937
00:50:34.900 --> 00:50:35.541
Does that make sense?

938
00:50:35.601 --> 00:50:35.821
Yeah.

939
00:50:36.461 --> 00:50:52.412
We're trying to, again, describe the acceleration or deceleration of an individual thing in an incompressible fluid from a vantage point that is only the vector field.

940
00:50:52.472 --> 00:50:52.733
Yes.

941
00:50:52.753 --> 00:50:53.353
Yes.

942
00:50:53.533 --> 00:51:02.758
Which the point being the vector field in and of itself does not have enough to describe the acceleration of the individual object until you do all of this math.

943
00:51:02.838 --> 00:51:03.598
Until you do all of this.

944
00:51:03.618 --> 00:51:04.579
Yeah.

945
00:51:04.659 --> 00:51:05.839
Is that the right way to say it back?

946
00:51:05.859 --> 00:51:07.060
I think that is the right way to say it.

947
00:51:07.080 --> 00:51:07.520
Okay.

948
00:51:07.580 --> 00:51:07.960
Exactly.

949
00:51:07.980 --> 00:51:09.401
And so now we have an acceleration.

950
00:51:09.901 --> 00:51:10.061
Right.

951
00:51:10.101 --> 00:51:10.261
Right?

952
00:51:10.542 --> 00:51:14.564
And all we need for that other part of Newton's laws, right, is F equals MA.

953
00:51:14.624 --> 00:51:16.244
We already have the A, the acceleration.

954
00:51:16.285 --> 00:51:16.445
Yeah.

955
00:51:16.485 --> 00:51:17.085
What is the mass?

956
00:51:17.125 --> 00:51:19.686
Well, the mass is just, for a fluid, it's like the density.

957
00:51:20.146 --> 00:51:20.327
Yeah.

958
00:51:20.427 --> 00:51:20.647
Right?

959
00:51:21.007 --> 00:51:22.668
It's just, I could do a...

960
00:51:23.428 --> 00:51:24.968
F equals ma per volume.

961
00:51:25.528 --> 00:51:28.269
And then just say, okay, density multiplied by acceleration.

962
00:51:28.729 --> 00:51:31.070
That's my, that part of Newton's laws.

963
00:51:31.090 --> 00:51:31.670
That's my ma.

964
00:51:31.750 --> 00:51:32.630
Yeah, that's my ma.

965
00:51:32.850 --> 00:51:33.990
And that's exactly what we get.

966
00:51:34.150 --> 00:51:36.411
We've got the left-hand side of Navier-Stokes equations.

967
00:51:36.511 --> 00:51:36.831
Okay.

968
00:51:36.871 --> 00:51:38.191
The first Navier-Stokes equation.

969
00:51:38.271 --> 00:51:38.451
Right.

970
00:51:38.691 --> 00:51:40.932
Because density multiplied by acceleration.

971
00:51:41.272 --> 00:51:41.472
Right.

972
00:51:42.012 --> 00:51:42.252
Right.

973
00:51:42.912 --> 00:51:46.193
Because our acceleration is everything in this larger bracket.

974
00:51:46.253 --> 00:51:46.453
Yeah.

975
00:51:47.573 --> 00:51:48.934
Which is what we just described.

976
00:51:49.174 --> 00:51:50.294
Yeah.

977
00:51:50.774 --> 00:51:54.956
Which is trying to describe the acceleration of an individual object from the helicopter.

978
00:51:55.276 --> 00:51:55.717
Exactly.

979
00:51:55.937 --> 00:51:57.217
Okay, got it.

980
00:51:57.237 --> 00:52:06.121
So now, again, we're trying to go, we're trying to use Newton's laws as a language to describe stuff that's happening in an incompressible fluid.

981
00:52:06.141 --> 00:52:07.362
Yes.

982
00:52:07.782 --> 00:52:16.286
And we now have half of Newton's force times mass equals acceleration for an incompressible fluid with the description that we've just gone through.

983
00:52:16.466 --> 00:52:16.886
Exactly.

984
00:52:17.166 --> 00:52:25.932
And now I want to, before we even start talking about the other side of that equation, which is the force, I want to linger on this acceleration term a little bit more, especially that convective term.

985
00:52:25.952 --> 00:52:30.514
Because that convective term, you might have heard that the Navier-Stokes equations are nonlinear.

986
00:52:30.895 --> 00:52:31.135
Yes.

987
00:52:31.215 --> 00:52:31.415
Right?

988
00:52:31.495 --> 00:52:33.256
Compared to linear differential equations.

989
00:52:33.576 --> 00:52:36.558
The Navier-Stokes equation is a nonlinear differential equation.

990
00:52:37.098 --> 00:52:39.000
That second term, the convective term there...

991
00:52:39.580 --> 00:52:40.861
has everything to do with that.

992
00:52:41.161 --> 00:52:46.344
Okay, so let's linger on that for a little bit longer because I want to show you exactly what all the trouble is.

993
00:52:46.805 --> 00:53:01.794
And just to reiterate here, the convective term that we have on the right was when we were talking about the car in the yellow or in the red moving from the 100 meters per second to the 80 meters per second, describing the acceleration in this slice of the larger system.

994
00:53:01.995 --> 00:53:02.795
Yeah, and exactly.

995
00:53:02.855 --> 00:53:05.097
And crucially, it had to deal with...

996
00:53:05.837 --> 00:53:10.842
its own velocity, and the spatial difference in the velocity, right?

997
00:53:10.922 --> 00:53:12.964
That's why it's nonlinear, because there's two.

998
00:53:13.304 --> 00:53:15.667
There's a compounding effect that is happening.

999
00:53:15.727 --> 00:53:16.127
Okay.

1000
00:53:16.147 --> 00:53:16.628
Okay?

1001
00:53:17.228 --> 00:53:21.052
Let's take a step back and let's look at, for example, Maxwell's equations in a vacuum.

1002
00:53:21.072 --> 00:53:21.252
Okay.

1003
00:53:21.292 --> 00:53:21.492
Okay?

1004
00:53:21.753 --> 00:53:24.055
These are the equations of electricity and magnetism.

1005
00:53:24.315 --> 00:53:24.435
Okay.

1006
00:53:24.455 --> 00:53:27.218
They're described, for example, Gauss's Law for Electric Fields...

1007
00:53:27.858 --> 00:53:34.002
is that the divergence of the electric field is equal to how much charge there is.

1008
00:53:34.082 --> 00:53:36.764
You know how before I was telling you that fluids are divergence-less?

1009
00:53:37.004 --> 00:53:37.204
Yes.

1010
00:53:37.244 --> 00:53:38.225
The divergence is zero.

1011
00:53:38.245 --> 00:53:42.628
But I showed you those two examples where the vector field was going out or the vector field was coming in.

1012
00:53:43.088 --> 00:53:45.410
And that's because you can have negative charge and positive charge, right?

1013
00:53:45.790 --> 00:53:47.631
That's Gauss's law for electricity, okay?

1014
00:53:47.871 --> 00:53:52.534
It just says that del.e, the divergence of the electric field, equals to some charge.

1015
00:53:52.915 --> 00:53:57.097
Now, crucially, the divergence of stuff is a linear operator.

1016
00:53:57.618 --> 00:54:00.840
Meaning, if I've got one electric field and another electric field,

1017
00:54:02.062 --> 00:54:11.286
and I want to find the divergence of the two of the electric fields put together, I can just find the divergence of one, and find the divergence of the other, and I can add them up separately.

1018
00:54:11.586 --> 00:54:19.329
So if I want to combine two electric fields, I just do the math for one of them, I do the math for the other one, and I combine them, and that's totally fine.

1019
00:54:19.849 --> 00:54:30.994
This is why, for example, if I've got two charges, a negative charge and a positive charge, and I want to find the electric field everywhere, all I do is find the electric field due to one, the electric field due to the other, and I add it up.

1020
00:54:31.749 --> 00:54:31.989
Okay.

1021
00:54:32.389 --> 00:54:36.552
This is also why telecommunications works with electric fields, right?

1022
00:54:36.592 --> 00:54:42.595
Like in this room, we've got Wi-Fi, we've got Bluetooth, we've got radio, we've got our phones.

1023
00:54:42.935 --> 00:54:45.817
We also have the light from the studio, right?

1024
00:54:46.297 --> 00:54:47.938
All of these are electric fields.

1025
00:54:48.853 --> 00:54:51.075
But they're all just moving through one another.

1026
00:54:51.115 --> 00:54:53.937
They're like ghosts that are just like moving through one another.

1027
00:54:54.217 --> 00:54:59.541
And if I can tune my radio to one thing or the other, I can listen to one station or the other.

1028
00:54:59.821 --> 00:55:05.145
Or I can, like, my Bluetooth isn't randomly, like, interfering with the Wi-Fi.

1029
00:55:05.165 --> 00:55:06.766
Like, the Wi-Fi is coming from the router.

1030
00:55:07.066 --> 00:55:09.167
There's a Bluetooth that's probably in my phone.

1031
00:55:09.428 --> 00:55:12.570
But it's not like the Bluetooth signal and the Wi-Fi signal is, like, crashing.

1032
00:55:13.490 --> 00:55:42.006
right yeah they don't interfere with each other yeah yeah even like um in a radio right in the car like if i'm tuned to 91.5 kusc classical radio um the car is being inundated with all of the radio stations there's also 89.9 coming at me and kiss fm and also the am you know talk radio rush not sponsored we need our check yeah these are just some of my favorites that i listen to in la but like

1033
00:55:43.658 --> 00:55:50.204
My point is the car is being inundated with all of these radio stations, but my radio can tune in into a specific band.

1034
00:55:50.564 --> 00:55:57.251
And it's not like the KISS FM radio signal is interfering with the KUSC signal, is interfering with Rush Limbaugh.

1035
00:55:57.612 --> 00:55:58.913
Or he's not...

1036
00:55:59.193 --> 00:56:00.054
I think he's still on radio.

1037
00:56:00.135 --> 00:56:00.916
Yeah, he's still on radio.

1038
00:56:01.076 --> 00:56:02.198
I think he's still on radio.

1039
00:56:02.218 --> 00:56:03.059
He might be on YouTube now.

1040
00:56:03.099 --> 00:56:04.000
Yeah, yeah, yeah.

1041
00:56:04.040 --> 00:56:04.621
Or a satellite.

1042
00:56:04.741 --> 00:56:09.528
But I think, so part of what you're saying is, like, I don't have to account for the other radio stations.

1043
00:56:09.548 --> 00:56:10.009
Exactly.

1044
00:56:10.310 --> 00:56:11.772
It doesn't matter what the other radio stations are.

1045
00:56:11.792 --> 00:56:13.715
And if I want to account for it, all I do is add it.

1046
00:56:14.703 --> 00:56:18.347
If I want to know what is the electric field here, all I do is add up all of the contributions.

1047
00:56:18.607 --> 00:56:20.049
It's totally fine.

1048
00:56:20.589 --> 00:56:23.011
In fluids, we do not have that luxury.

1049
00:56:23.492 --> 00:56:23.772
Okay.

1050
00:56:24.193 --> 00:56:26.135
And that is because of that convective term.

1051
00:56:26.975 --> 00:56:28.597
That convective term is nonlinear.

1052
00:56:28.677 --> 00:56:32.921
And this is what it does if I add up two velocity fields that are separate.

1053
00:56:33.021 --> 00:56:36.605
Suppose, you know, and before I was adding up E1 and E2, the electric fields.

1054
00:56:36.645 --> 00:56:36.885
Yes.

1055
00:56:37.105 --> 00:56:37.766
It's totally fine.

1056
00:56:37.786 --> 00:56:38.026
Fine.

1057
00:56:38.046 --> 00:56:38.146
Fine.

1058
00:56:38.827 --> 00:56:40.888
Suppose I have two velocity fields, right?

1059
00:56:40.928 --> 00:56:43.709
I got two hoses and I'm like crossing streams or something.

1060
00:56:43.830 --> 00:56:44.710
U1 plus U2.

1061
00:56:44.730 --> 00:56:46.471
I shouldn't have, you know what?

1062
00:56:46.711 --> 00:56:47.031
Whatever.

1063
00:56:47.251 --> 00:56:48.852
Like two things of water.

1064
00:56:49.552 --> 00:56:51.153
And I add up, right?

1065
00:56:51.173 --> 00:56:51.854
Like U1 and U2.

1066
00:56:53.715 --> 00:56:56.376
Suppose I go through the math with those.

1067
00:56:57.037 --> 00:56:59.958
If it was linear, I would only get the terms in the green.

1068
00:57:00.838 --> 00:57:01.038
Right?

1069
00:57:01.098 --> 00:57:06.301
Because I could just do the math for one, U1, and I could do the math for U2, and I'd be fine.

1070
00:57:06.961 --> 00:57:08.402
You do the divergence for U1.

1071
00:57:08.462 --> 00:57:12.103
Yeah, divergence for U2 multiplied by U2 and U1, and that'd be it.

1072
00:57:12.223 --> 00:57:12.504
Okay.

1073
00:57:12.564 --> 00:57:13.324
But because...

1074
00:57:14.646 --> 00:57:18.347
It's u multiplied by the divergence of itself, right?

1075
00:57:18.427 --> 00:57:20.588
Or it's multiplied by the gradient, I should say, of itself.

1076
00:57:20.628 --> 00:57:22.889
Not divergence because divergence has a dot.

1077
00:57:23.049 --> 00:57:28.731
Anyways, the point is I'm taking the velocity field itself and I'm multiplying by the derivative of itself.

1078
00:57:29.371 --> 00:57:32.352
Because I'm doing that, I have to do the distributive property.

1079
00:57:33.072 --> 00:57:34.013
You have to do it twice.

1080
00:57:34.353 --> 00:57:35.313
Yeah, I have to do it twice.

1081
00:57:35.373 --> 00:57:38.374
And so I'm going to get these cross terms on the right in the red circle.

1082
00:57:38.514 --> 00:57:38.834
Okay.

1083
00:57:38.994 --> 00:57:40.355
And those cross terms...

1084
00:57:41.562 --> 00:57:43.543
are what make this whole thing nonlinear.

1085
00:57:44.283 --> 00:57:47.165
And that gives us all of the richness of fluids.

1086
00:57:47.205 --> 00:57:49.886
This is why fluids don't just go through one another.

1087
00:57:50.227 --> 00:57:51.367
They crash.

1088
00:57:51.407 --> 00:57:53.208
There's rapids, right?

1089
00:57:53.428 --> 00:57:58.691
When two rivers meet together, they mix and they have turbulent flow.

1090
00:57:59.872 --> 00:58:06.135
When I was in India last time, we visited Devprayag in the Himalayas, which is the birthplace of the Ganga.

1091
00:58:06.455 --> 00:58:10.761
where like you have like Bhagirathi coming from one side and Alakananda coming from the other side.

1092
00:58:10.781 --> 00:58:11.663
There are two rivers

1093
00:58:12.119 --> 00:58:17.043
from the Himalayas that meet in this very holy place in the Himalayas that begins the Ganges River.

1094
00:58:18.284 --> 00:58:21.667
And you could see the whirlpools forming right at that spot.

1095
00:58:21.947 --> 00:58:22.968
It was amazing.

1096
00:58:23.148 --> 00:58:31.355
And the reason for the whirlpools and the reason for all of the chaos in the birthplace of the Ganga is because of those two terms on the right-hand side.

1097
00:58:31.875 --> 00:58:32.436
You know what I mean?

1098
00:58:32.996 --> 00:58:33.176
Yeah.

1099
00:58:33.376 --> 00:58:37.500
And so now you can already see that there is trouble with this equation.

1100
00:58:38.240 --> 00:58:38.481
Right.

1101
00:58:38.721 --> 00:58:43.866
And I just want to take a quick step back to make sure I'm internalizing what we're saying here.

1102
00:58:45.747 --> 00:58:58.679
When we talked about Maxwell's equations and we have a positive and a negative charge and we can just add them up, it's because – part of what we're saying is it's because we don't have –

1103
00:58:59.500 --> 00:59:05.087
That extra level of complexity when we talked about the convective term that's true for fluids.

1104
00:59:05.148 --> 00:59:05.608
Yeah, yeah.

1105
00:59:05.708 --> 00:59:07.871
In the Maxwell's equation, it's just del dot e, right?

1106
00:59:07.931 --> 00:59:09.734
It's a single e. Right.

1107
00:59:09.754 --> 00:59:12.457
Here, there's two versions of the vector field.

1108
00:59:12.477 --> 00:59:15.962
There's the vector field itself multiplied by the derivative of the vector field.

1109
00:59:15.982 --> 00:59:16.402
The vector field.

1110
00:59:17.183 --> 00:59:17.403
Right.

1111
00:59:17.824 --> 00:59:21.166
And can we just briefly try to connect it back?

1112
00:59:21.826 --> 00:59:28.911
I'm trying to connect it back to our yellow-red traffic example just to make sure it lands in my brain.

1113
00:59:28.931 --> 00:59:34.895
Because I'm having a hard time finding the leap from – I understood the convective term.

1114
00:59:36.316 --> 00:59:41.159
We have this idea that the traffic is red in one part, yellow in one part.

1115
00:59:41.179 --> 00:59:42.580
And we can look at it from the helicopter.

1116
00:59:42.720 --> 00:59:44.081
And we can look at the whole vector field.

1117
00:59:46.965 --> 00:59:47.945
Or from red to green?

1118
00:59:48.458 --> 00:59:52.281
then it is going to accelerate, even though red and green have stayed put.

1119
00:59:53.062 --> 01:00:01.469
And so from that perspective, would it be like the equivalent of saying, if that was it and that was the only level of complexity, it would be similar to Maxwell's equations in that it's not there?

1120
01:00:01.489 --> 01:00:10.456
Yeah, if all I had to do was take a spatial derivative of red and green and be like, oh, there's red here and green here, and they're 100 meters apart.

1121
01:00:10.737 --> 01:00:13.419
So the difference in the velocity is divided by 100 meters.

1122
01:00:13.459 --> 01:00:16.481
That tells me the spatial derivative of my velocity, and I'm fine.

1123
01:00:16.641 --> 01:00:17.062
That's fine.

1124
01:00:17.082 --> 01:00:17.322
That's fine.

1125
01:00:17.402 --> 01:00:21.345
But crucially, it also has to do with how fast or how slow I was going into the thing.

1126
01:00:21.565 --> 01:00:21.805
Right?

1127
01:00:22.045 --> 01:00:30.892
If I'm transitioning from red to green and I'm moving at a snail's pace of like five miles an hour and all of a sudden now I have to get up to highway speed, I got to floor it.

1128
01:00:30.992 --> 01:00:31.172
Yes.

1129
01:00:31.352 --> 01:00:37.157
But if in the red I was already going at 30 and then now I got to get to 60, I don't have to floor it as hard.

1130
01:00:37.177 --> 01:00:37.657
Right.

1131
01:00:37.797 --> 01:00:39.418
So there's two contributions.

1132
01:00:39.758 --> 01:00:39.999
Right.

1133
01:00:40.119 --> 01:00:42.741
And that's what's giving us this non-linearity.

1134
01:00:42.841 --> 01:00:43.081
Yes.

1135
01:00:43.301 --> 01:00:45.062
The fact that there's two contributions.

1136
01:00:45.102 --> 01:00:50.406
Two contributions that we have to account for as we look at trying to do the mathematics for this.

1137
01:00:50.686 --> 01:00:54.309
And that's where we're getting – this is where we're getting having to do it –

1138
01:00:55.609 --> 01:00:59.671
We have the green, but we have to have the addition of the red because of that second.

1139
01:00:59.851 --> 01:01:01.632
Because of the fact that, yeah, there's two of them.

1140
01:01:01.792 --> 01:01:01.972
Right.

1141
01:01:01.992 --> 01:01:03.333
So I have to do these cross terms.

1142
01:01:03.813 --> 01:01:04.153
Understood.

1143
01:01:04.173 --> 01:01:04.333
Yeah.

1144
01:01:04.353 --> 01:01:04.533
Okay.

1145
01:01:04.573 --> 01:01:12.197
When I combine velocity fields, when I combine, let's say, a flow in this direction and a flow in this direction, they don't just ghost past one another.

1146
01:01:12.297 --> 01:01:12.457
Right.

1147
01:01:12.477 --> 01:01:13.517
Because they're interacting.

1148
01:01:13.537 --> 01:01:13.697
Yeah.

1149
01:01:13.937 --> 01:01:14.137
Yep.

1150
01:01:14.297 --> 01:01:14.538
Okay.

1151
01:01:14.798 --> 01:01:15.158
Make sense?

1152
01:01:15.178 --> 01:01:15.378
Yes.

1153
01:01:15.538 --> 01:01:15.898
Tracking.

1154
01:01:15.938 --> 01:01:16.178
Yeah.

1155
01:01:16.398 --> 01:01:17.179
And so that's...

1156
01:01:18.019 --> 01:01:20.500
why the Navier-Stokes are so hard.

1157
01:01:21.220 --> 01:01:22.701
That's one of the reasons.

1158
01:01:23.461 --> 01:01:26.803
So now let's get back to our Navier-Stokes equations.

1159
01:01:27.543 --> 01:01:31.885
Finally, we can now start asking, what is the force that is acting on the fluid?

1160
01:01:32.165 --> 01:01:38.008
So far, we've figured out how to describe the acceleration, and we've figured out how to describe the mass.

1161
01:01:38.668 --> 01:01:38.888
Right?

1162
01:01:39.869 --> 01:01:40.089
Yes.

1163
01:01:40.149 --> 01:01:45.651
But in order to have an equation of motion, we have to describe the forces that cause this acceleration.

1164
01:01:45.971 --> 01:01:49.072
And then if we write it out, then perhaps we can solve it.

1165
01:01:49.672 --> 01:01:49.832
Right?

1166
01:01:50.053 --> 01:01:54.274
In the same way that for the spring, we could solve mass times the acceleration.

1167
01:01:54.294 --> 01:01:55.895
In that case, the acceleration was super easy.

1168
01:01:55.915 --> 01:01:57.415
It's just the second derivative of the position.

1169
01:01:57.856 --> 01:01:59.376
And then the force was negative kx.

1170
01:01:59.736 --> 01:02:01.357
I had an equation of motion that I could solve.

1171
01:02:01.437 --> 01:02:04.858
In this case, already the acceleration is quite difficult.

1172
01:02:04.878 --> 01:02:05.479
Right?

1173
01:02:05.499 --> 01:02:05.639
Right?

1174
01:02:06.599 --> 01:02:08.580
But now let's ask, what is the force?

1175
01:02:09.000 --> 01:02:11.061
One obvious thing is pressure.

1176
01:02:12.602 --> 01:02:14.463
Pressure is force per unit area.

1177
01:02:15.563 --> 01:02:19.665
You know, higher pressure means that there's more sort of forces acting on

1178
01:02:21.185 --> 01:02:22.146
acting on your container.

1179
01:02:22.426 --> 01:02:31.011
Like, you know, if you've got a high-pressured gas inside of a container, that gas is pushing out on the walls of the container.

1180
01:02:31.351 --> 01:02:33.692
Compared to a low-pressure gas, it's not pushing out as much.

1181
01:02:33.712 --> 01:02:34.213
It's just hanging out.

1182
01:02:34.233 --> 01:02:34.893
Yeah, exactly.

1183
01:02:34.913 --> 01:02:44.758
Like when you suck on a straw, what you're really doing is creating low pressure in your mouth, and then the high pressure of the atmosphere pushes down on the liquid and forces it up into your mouth.

1184
01:02:44.978 --> 01:02:46.619
That's a difference in pressure that you're doing.

1185
01:02:46.860 --> 01:02:49.021
So crucially, what matters here, what is the force?

1186
01:02:49.938 --> 01:02:55.421
on the fluid, it matters what the difference in pressure is, not what the absolute pressure is.

1187
01:02:55.601 --> 01:03:03.366
There's absolute high, like a thousand, a hundred thousand pounds per square inch, right here, in this room.

1188
01:03:03.906 --> 01:03:06.828
But I don't feel it, because there's no pressure gradient.

1189
01:03:07.128 --> 01:03:09.009
There's no difference in pressure.

1190
01:03:09.029 --> 01:03:10.310
So I don't feel the force.

1191
01:03:10.390 --> 01:03:17.494
Now all of a sudden, if outside there was super low pressure for some reason, and we had like a leaf blower in here, and,

1192
01:03:19.135 --> 01:03:22.818
There would be high pressure in the room, low pressure outside, and we'd feel a wind that's going out.

1193
01:03:23.138 --> 01:03:24.819
In weather systems, you see this a lot, right?

1194
01:03:24.839 --> 01:03:28.982
High pressure, the wind goes away, goes away from the high pressure towards the low pressure.

1195
01:03:29.243 --> 01:03:33.185
So one of the forces that we will have to worry about is the gradient in the pressure.

1196
01:03:33.646 --> 01:03:36.148
How the pressure changes from one space to another.

1197
01:03:36.668 --> 01:03:38.830
And it's negative because it's going from high to low.

1198
01:03:39.030 --> 01:03:40.711
So the force is always from high to low.

1199
01:03:40.751 --> 01:03:41.131
Right, right.

1200
01:03:41.211 --> 01:03:41.452
Okay.

1201
01:03:41.592 --> 01:03:47.256
Because it's going from a high, it's always flowing from an environment of high pressure to an environment of low pressure.

1202
01:03:47.556 --> 01:03:50.817
So you can always have the negative in front of it because it's always going to be.

1203
01:03:50.837 --> 01:03:52.478
Yeah, it's always going to go from high to low pressure.

1204
01:03:52.758 --> 01:03:54.298
And so this is a force, right?

1205
01:03:54.398 --> 01:03:58.520
The whole point right now is we're trying to describe what are the forces on the fluid.

1206
01:03:58.960 --> 01:04:04.682
In this case, the forces on that velocity field are from negative to from high pressure to low pressure.

1207
01:04:04.722 --> 01:04:05.182
You're going to have.

1208
01:04:06.170 --> 01:04:33.738
velocity is flowing right it's like if you're at a beach and you know those little videos where the surfers create oh yeah yeah the the surfing thing on the from a beach into the ocean yeah and it's it's like the the water pressure is very high and built up at the top and then it wants to flow into the ocean yeah which is a relative area of lower pressure yeah in that context in that context actually you're the next one that i want to talk about is the external forces on my fluid

1209
01:04:36.079 --> 01:04:36.259
Right.

1210
01:04:36.279 --> 01:04:41.180
Because the the like the lagoon or whatever is slightly above the ocean.

1211
01:04:41.200 --> 01:04:41.680
That's a good point.

1212
01:04:41.700 --> 01:04:41.841
Right.

1213
01:04:41.901 --> 01:04:48.642
And so when you create that little tiny thing, gravity is going to force the fluid down that little canyon that you've built.

1214
01:04:48.662 --> 01:04:53.704
So my example was actually it was not perfectly apt because I was not yet accounting for gravity.

1215
01:04:53.724 --> 01:04:53.964
Yes.

1216
01:04:54.064 --> 01:04:58.205
And the next thing that we're going to do is account for external forces of the fluid.

1217
01:04:58.345 --> 01:04:58.525
OK.

1218
01:04:59.613 --> 01:05:01.074
like gravity, right?

1219
01:05:01.094 --> 01:05:02.095
That makes sense.

1220
01:05:02.115 --> 01:05:03.857
So on the right-hand side, we've got forces.

1221
01:05:04.117 --> 01:05:04.317
Yes.

1222
01:05:05.037 --> 01:05:07.059
The difference in pressure is one such force.

1223
01:05:07.139 --> 01:05:07.359
Yes.

1224
01:05:07.459 --> 01:05:21.071
That is going to, you know, like if I were to increase the pressure here and decrease the pressure here, and I started out with some velocity field, those vectors are going to get bigger because there's going to be more fluid going through and the fluid is going to be moving faster.

1225
01:05:21.371 --> 01:05:23.353
Similarly, if I introduce gravity now,

1226
01:05:24.217 --> 01:05:25.899
the velocity is going to get bigger.

1227
01:05:26.119 --> 01:05:27.240
So that's a kind of force.

1228
01:05:27.560 --> 01:05:30.082
It's causing an acceleration, right?

1229
01:05:30.583 --> 01:05:34.646
And so here we've already got somewhat of a Navier-Stokes equation, right?

1230
01:05:35.026 --> 01:05:39.871
We've got a familiar example is gravity, but honestly that g could be a lot of things.

1231
01:05:40.591 --> 01:05:43.614
That g could be the stirring force, right?

1232
01:05:44.687 --> 01:05:44.847
Right?

1233
01:05:44.887 --> 01:05:52.972
Like if I put a spoon in a piece of, in a glass and I stir the thing, that's an external force that's acting on my fluid.

1234
01:05:53.352 --> 01:05:55.173
So the force can be a lot of things.

1235
01:05:55.694 --> 01:05:56.154
Understood.

1236
01:05:56.494 --> 01:05:59.156
It's not necessarily just gravity in the context.

1237
01:05:59.176 --> 01:05:59.976
We just talked about it.

1238
01:06:00.397 --> 01:06:00.597
Yeah.

1239
01:06:00.617 --> 01:06:02.258
We just talked about gravity in that context.

1240
01:06:02.298 --> 01:06:04.499
But I mean, forces on fluids can be a lot of things.

1241
01:06:04.539 --> 01:06:07.681
It can be like the turbine in an airplane.

1242
01:06:07.721 --> 01:06:09.942
That's putting a force on, right?

1243
01:06:09.962 --> 01:06:13.645
Because the thing is, it's stirring a fluid and making it shoot out in one direction.

1244
01:06:13.685 --> 01:06:14.245
Yeah, yeah, yeah.

1245
01:06:15.339 --> 01:06:21.663
So that is about as far as one of our favorite mathematicians on this podcast, Leonard Euler, got.

1246
01:06:22.203 --> 01:06:27.986
And I think this is a good point to talk about some of the history behind the Navier-Stokes equation.

1247
01:06:28.006 --> 01:06:34.090
Because that last equation that we had is one of the Euler equations of fluids.

1248
01:06:34.210 --> 01:06:39.012
And just to quickly kind of recap, we've started, we're trying to describe...

1249
01:06:41.835 --> 01:06:55.119
The acceleration of an individual entity within a vector field from only having the context of the vector field.

1250
01:06:56.939 --> 01:07:00.080
And we are trying to describe it in the language of Newton's second law.

1251
01:07:01.060 --> 01:07:03.761
Force equals mass times acceleration.

1252
01:07:03.961 --> 01:07:04.161
Yeah.

1253
01:07:04.521 --> 01:07:08.142
And we've built up now for an incompressible fluid.

1254
01:07:08.282 --> 01:07:08.462
Yeah.

1255
01:07:10.235 --> 01:07:15.580
All the way up to where Euler arrived in this journey of trying to do so.

1256
01:07:15.780 --> 01:07:16.040
Yes.

1257
01:07:16.460 --> 01:07:18.122
He didn't get there on his own, though.

1258
01:07:18.382 --> 01:07:18.702
Oh, okay.

1259
01:07:19.083 --> 01:07:19.343
Ooh.

1260
01:07:19.583 --> 01:07:19.843
Okay.

1261
01:07:20.003 --> 01:07:24.187
We said no drama till the end, but... And there is some drama here.

1262
01:07:24.227 --> 01:07:28.611
There's actually some really weird drama that is very pertinent to today's drama.

1263
01:07:28.631 --> 01:07:28.891
Yeah.

1264
01:07:29.211 --> 01:07:31.252
Of OpenAI versus all these mathematicians.

1265
01:07:31.272 --> 01:07:33.793
So I thought it was kind of interesting to talk about.

1266
01:07:33.873 --> 01:07:37.254
So Leonard Euler, you know, the great Leonard Euler, we've talked about him a lot.

1267
01:07:37.294 --> 01:07:47.499
In the Riemann Zeta episode, we talked about how Leonard Euler invented the Zeta function that then Riemann later extended to create the Riemann Zeta function.

1268
01:07:48.379 --> 01:07:52.240
function and the Riemann hypothesis, it's a very similar story here.

1269
01:07:52.700 --> 01:07:59.383
Euler created the Euler equations, and then Navier and Stokes extended those equations to create the Navier-Stokes equations that we know today.

1270
01:07:59.863 --> 01:08:06.105
And obviously, you want to name things after the second person who invented everything, because you can't name everything after Euler.

1271
01:08:06.909 --> 01:08:10.213
But even Euler inventing that first equation, that wasn't on his own.

1272
01:08:10.513 --> 01:08:16.840
Okay, there is a lot of shenanigans that happened in 1940s Berlin between... 1740s.

1273
01:08:17.381 --> 01:08:18.903
Sorry, yeah, of course.

1274
01:08:18.983 --> 01:08:22.067
Yeah, 1740s Berlin between Leonard Euler.

1275
01:08:22.207 --> 01:08:27.192
He was working at the Royal Prussian Academy of Scientists.

1276
01:08:27.813 --> 01:08:30.274
This is the German Academy of Sciences, okay?

1277
01:08:30.494 --> 01:08:34.235
And if you go to Berlin, you can find a little plaque for where Leonard Euler used to live.

1278
01:08:34.855 --> 01:08:43.517
So the German Academy of Sciences announces a prize problem about one of the great practical questions in the 1740s.

1279
01:08:44.817 --> 01:08:50.399
How much resistance does a body experience as it moves through air or water?

1280
01:08:51.023 --> 01:08:51.263
Okay.

1281
01:08:51.423 --> 01:08:56.989
It's a very pressing question because I think around the time is when cannons are being used in warfare.

1282
01:08:57.249 --> 01:09:00.852
So it's like, you know, we want to know how far a cannon can go.

1283
01:09:00.972 --> 01:09:09.140
It's their version of the DoD creating, you know, a prize competition to try to understand how to do better warfare.

1284
01:09:09.861 --> 01:09:13.424
A guy named Jean Leronde Dupin.

1285
01:09:13.504 --> 01:09:14.465
D'Alembert.

1286
01:09:14.945 --> 01:09:15.445
Very good.

1287
01:09:15.585 --> 01:09:16.986
For those who are long-time listeners.

1288
01:09:17.926 --> 01:09:20.908
I don't usually do this, but I like D'Alembert.

1289
01:09:20.928 --> 01:09:22.949
So he submits an essay.

1290
01:09:24.097 --> 01:09:25.598
It's about fluids.

1291
01:09:25.938 --> 01:09:38.463
He did not solve the resistance problem completely, but he helped introduce something that is arguably maybe a lot more important, which is this mathematical language of a velocity field to describe a fluid.

1292
01:09:38.784 --> 01:09:44.026
He's the guy who invented that concept of U, of the velocity field.

1293
01:09:44.046 --> 01:09:52.750
You know, all this stuff that we've been talking about, this new language of maybe let's not worry about all of the particles, let's worry about the velocity field of this thing, that was D'Alembert.

1294
01:09:53.605 --> 01:10:08.786
And, okay, so what's interesting is I want to make a quick zoom in note to note that the prize problem that he set out to answer was about how much resistance does a body experience when it moves through air or water.

1295
01:10:11.057 --> 01:10:14.180
And I just want to be, that's the specific question he was trying to answer.

1296
01:10:14.940 --> 01:10:17.843
In that journey, he did not answer that question.

1297
01:10:17.883 --> 01:10:19.684
He did not answer the question, but he invented.

1298
01:10:19.704 --> 01:10:20.845
He gave us the language.

1299
01:10:20.905 --> 01:10:22.386
He gave us the language for it, right?

1300
01:10:22.586 --> 01:10:28.511
And his submission didn't win because the Academy of Sciences was like, oh, it's too theoretical.

1301
01:10:28.551 --> 01:10:33.755
He also showed that the velocity field has to be divergence-less.

1302
01:10:34.315 --> 01:10:38.396
meaning that del.u equals zero, like the fact that it's incompressible and things like that.

1303
01:10:38.676 --> 01:10:42.978
So he already established one of the two Navier-Stokes equations, Dal and Baird.

1304
01:10:43.018 --> 01:10:43.858
He submits this thing.

1305
01:10:43.878 --> 01:10:46.499
The Academy decides that it's not going to give out the prize.

1306
01:10:47.299 --> 01:10:48.259
No prize was awarded.

1307
01:10:48.800 --> 01:10:49.500
But here's the thing.

1308
01:10:49.840 --> 01:10:52.841
One of the people that was on the committee was Leonard Euler.

1309
01:10:54.174 --> 01:11:03.426
So Euler saw an early submission of D'Alembert's work, and he's reading this thing going, yo, this velocity field is a kind of nice idea.

1310
01:11:04.336 --> 01:11:13.619
He takes that idea and not long afterward, he presents his own general theory of incompressible fluid motion and that's the equation that we see on the right-hand side.

1311
01:11:13.639 --> 01:11:24.623
So on the left-hand side, Dahlembert has already invented the velocity field and he's shown that the velocity field has to have zero divergence, meaning it's incompressible, no sources or sinks.

1312
01:11:25.043 --> 01:11:29.465
Leonard Euler takes that concept and says, well, I can actually make a Newton's Law version

1313
01:11:30.678 --> 01:11:35.300
So very similar to Riemann zeta where the Basel problem existed beforehand, right?

1314
01:11:35.440 --> 01:11:35.860
Right.

1315
01:11:35.880 --> 01:11:39.662
The 1 plus 1 over 4 plus 1 over 9 plus 1 over 16, so on and so forth.

1316
01:11:40.282 --> 01:11:46.085
Leonard Euler saw the Basel problem, solved it, and then presented the zeta function.

1317
01:11:46.405 --> 01:11:51.267
In this case, he saw D'Alembert's velocity field and he's like, wait, this is a good idea.

1318
01:11:51.367 --> 01:11:56.290
And he just ran with it and created the derivatives, which are huge, like the...

1319
01:11:57.833 --> 01:12:03.115
The derivative term, that acceleration, is massive to create that material derivative.

1320
01:12:03.416 --> 01:12:09.098
And then also say that, okay, the forces on it are going to be the gradient in the pressure plus whatever external force.

1321
01:12:09.358 --> 01:12:13.160
And just to say what you just spoke to, that was the convective term that we talked about earlier.

1322
01:12:13.180 --> 01:12:14.480
Yeah, yeah.

1323
01:12:14.921 --> 01:12:15.181
Yes.

1324
01:12:15.781 --> 01:12:15.961
Okay.

1325
01:12:16.021 --> 01:12:17.942
And this is so interesting because it's...

1326
01:12:20.938 --> 01:12:30.580
It's always funny how these things... Basically, we have D'Alembert, who created the vision to think about it this way.

1327
01:12:31.120 --> 01:12:39.401
And then Euler was like, let me take Newton's framework, D'Alembert's vision, and combine these two things.

1328
01:12:39.561 --> 01:12:39.801
Yes.

1329
01:12:40.181 --> 01:12:42.322
And he gets an equation of motion, right?

1330
01:12:42.562 --> 01:12:44.602
He gets the first equation of motion for a fluid.

1331
01:12:44.942 --> 01:12:47.583
This is for an inviscid fluid, as they say.

1332
01:12:47.923 --> 01:12:48.603
Because these are...

1333
01:12:49.523 --> 01:12:51.224
as we'll find out, frictionless fluids.

1334
01:12:51.825 --> 01:12:52.065
Okay?

1335
01:12:52.345 --> 01:12:54.327
Now, obviously, that can't be everything.

1336
01:12:54.427 --> 01:12:57.529
And D'Alembert sees this.

1337
01:12:57.649 --> 01:13:00.992
It's kind of like a mathematical twist that's so perfect it's got to be scripted.

1338
01:13:01.432 --> 01:13:05.455
D'Alembert looks at the original competition, and he's like, wait a minute, it was about drag.

1339
01:13:06.276 --> 01:13:12.841
If I apply Newton's, I mean, sorry, if I apply Euler's equation to the concept of drag, and I ask, what is the drag on a body?

1340
01:13:13.201 --> 01:13:17.725
I can exactly solve Euler's equations around, let's say, a ball moving through a fluid.

1341
01:13:18.975 --> 01:13:20.056
And there's no drag.

1342
01:13:20.536 --> 01:13:25.681
Like in the next thing we'll see, this is called D'Alembert's paradox.

1343
01:13:26.081 --> 01:13:26.341
Okay.

1344
01:13:26.562 --> 01:13:32.367
He uses Euler's equation that's up there and he applies it to a ball, a sphere that is moving through fluid.

1345
01:13:34.088 --> 01:13:39.672
According to Euler's equations, if you were to solve with all those boundary conditions, the streamlines just go up and then they come back.

1346
01:13:40.372 --> 01:13:43.975
There's no difference in pressure from the front to the back.

1347
01:13:44.395 --> 01:13:45.396
And so there's no drag.

1348
01:13:45.436 --> 01:13:54.542
But obviously, as we know from our FIFA World Cup episode, where we talked about the drag on a soccer ball, there is a lot of drag on a soccer ball.

1349
01:13:55.362 --> 01:14:03.184
Just to also note, this was Adidas' peer-reviewed study on their new balls for this year's World Cup.

1350
01:14:03.384 --> 01:14:08.266
And they did incredible deep dive studies into the literal drag of how they designed it.

1351
01:14:08.286 --> 01:14:09.386
It's a great episode.

1352
01:14:09.406 --> 01:14:09.906
Definitely go watch it.

1353
01:14:09.946 --> 01:14:10.346
Exactly.

1354
01:14:10.406 --> 01:14:14.607
And what we can see already is in Real Fluids, the streamlines aren't just meeting up right after.

1355
01:14:14.667 --> 01:14:15.327
Yeah.

1356
01:14:15.667 --> 01:14:16.148
Right?

1357
01:14:16.168 --> 01:14:19.849
There's this region where there's turbulence and all this weird stuff that's happening.

1358
01:14:19.869 --> 01:14:20.669
This chaos happening.

1359
01:14:20.689 --> 01:14:20.829
Right?

1360
01:14:20.849 --> 01:14:21.549
Right?

1361
01:14:21.809 --> 01:14:23.750
Euler's equation does not account for that.

1362
01:14:23.770 --> 01:14:33.657
So basically what we're saying is Euler's equations only accounts for what we see on the left, which is effectively a system that's not actually exists in real context.

1363
01:14:33.717 --> 01:14:34.077
Exactly.

1364
01:14:34.097 --> 01:14:35.138
The military can't use it.

1365
01:14:35.198 --> 01:14:35.878
Exactly, yeah.

1366
01:14:36.098 --> 01:14:41.102
And so 20 years after the whole competition and everything, Dahlembert says, this is a paradox.

1367
01:14:41.582 --> 01:14:42.502
This is not everything.

1368
01:14:42.523 --> 01:14:43.323
There's got to be more.

1369
01:14:43.843 --> 01:14:45.865
And this is the part that Navier and Stokes...

1370
01:14:46.675 --> 01:14:47.456
Help us repair.

1371
01:14:48.096 --> 01:14:48.376
Okay?

1372
01:14:48.717 --> 01:14:52.720
They're the ones that come in and they start accounting for viscosity.

1373
01:14:54.001 --> 01:15:03.569
And so to now kind of quickly recap again, we are trying to build this understanding of what is the construction of the Navier-Stokes equation.

1374
01:15:03.930 --> 01:15:08.313
We understand d'Alembert started with this idea that the divergence equals zero.

1375
01:15:08.333 --> 01:15:08.434
Yeah.

1376
01:15:09.294 --> 01:15:34.648
uh oiler has come in and used the framework of newton's second law of motion to try to apply it to fluids maybe cheating off of d'alembert's test maybe a little bit maybe yeah um and added for the force side of the equation uh pressure uh and just an external and it's and an external gravity or stirring or whatever whatever it might be yeah and he was like i did it yeah look at me i'm so brilliant yeah however

1377
01:15:35.828 --> 01:15:44.273
When applied in practice to a ball like we just described, it does not actually describe real world systems.

1378
01:15:45.114 --> 01:15:52.819
And what's interesting is the – it's funny because the original prize competition that Euler was a judge on was about resistance and drag.

1379
01:15:52.919 --> 01:15:56.481
And then he came up with a solution without thinking about resistance and drag.

1380
01:15:56.501 --> 01:15:56.601
Yeah.

1381
01:15:56.941 --> 01:15:57.881
Am I missing something?

1382
01:15:57.941 --> 01:15:58.361
Exactly.

1383
01:15:58.441 --> 01:15:59.622
Okay.

1384
01:16:00.022 --> 01:16:04.803
So what he ended up missing was another force that he hadn't accounted for.

1385
01:16:04.883 --> 01:16:06.323
Remember, right now we have two forces.

1386
01:16:06.343 --> 01:16:07.724
We've got the difference in pressure.

1387
01:16:07.864 --> 01:16:08.064
Yes.

1388
01:16:08.284 --> 01:16:11.544
And we've got just external forces, whatever they might be, stirring, gravity, whatever.

1389
01:16:11.865 --> 01:16:11.965
Sure.

1390
01:16:11.985 --> 01:16:17.906
There is a third force on that side of the equation that is missing, and that has to do with viscosity.

1391
01:16:18.346 --> 01:16:18.606
Okay.

1392
01:16:18.926 --> 01:16:21.127
That's the resolution to D'Alembert's paradox.

1393
01:16:21.207 --> 01:16:21.467
Okay.

1394
01:16:21.487 --> 01:16:24.368
So in Euler's ideal fluid, there is no viscosity.

1395
01:16:24.408 --> 01:16:26.388
This is called an inviscid flow.

1396
01:16:27.849 --> 01:16:30.129
The fluids can slide past one another.

1397
01:16:30.389 --> 01:16:37.632
These particles can just slide past one another, and they can slide along the surface of an object without any friction.

1398
01:16:38.312 --> 01:16:42.833
And the flow divides very smoothly in front of the object, and then it rejoins at the end of the object.

1399
01:16:43.133 --> 01:16:43.894
There's no wake.

1400
01:16:44.315 --> 01:16:45.456
There's no net resistance.

1401
01:16:45.496 --> 01:16:46.318
There's no draft.

1402
01:16:46.538 --> 01:16:47.799
And this is beautiful mathematics.

1403
01:16:47.860 --> 01:16:51.384
It's different than diverge.

1404
01:16:51.404 --> 01:16:55.810
Like it's different than the problem we were saying where water –

1405
01:16:57.662 --> 01:17:20.997
form a incompressibility right right yeah this is different it's different this is still incompressible right because the flow is coming in and then it's going out it's just that they when they interact with each other it just rolls yeah they're just they just roll off yeah okay yeah they're just rolling off one another it's not especially useful okay we haven't taken into account friction and we have to take into account something called viscosity viscosity is simply the measure of how thick

1406
01:17:21.717 --> 01:17:25.358
sticky and resistant to flowing a liquid is.

1407
01:17:25.598 --> 01:17:30.259
So we're going from thick with two C's on the left here to thick with three C's on the right.

1408
01:17:30.359 --> 01:17:30.639
Yes.

1409
01:17:30.920 --> 01:17:31.220
Okay.

1410
01:17:31.820 --> 01:17:32.020
Yeah.

1411
01:17:32.300 --> 01:17:34.240
Vegetable oil, not that thick, right?

1412
01:17:34.260 --> 01:17:36.161
You drop something, it immediately goes through.

1413
01:17:36.861 --> 01:17:39.782
Motor oil, also not that thick, but slightly a little bit thicker.

1414
01:17:40.562 --> 01:17:48.464
Honey, on the other hand, you drop something and it takes forever to get all the way down because there's so much resistance to flow in honey.

1415
01:17:49.084 --> 01:17:50.865
The things as they try to move past each other

1416
01:17:51.942 --> 01:17:52.483
Make it harder.

1417
01:17:52.583 --> 01:17:53.304
Exactly, yeah.

1418
01:17:53.404 --> 01:18:00.270
And so for low viscosity, like thin liquids, like oils, the particles are moving past each other with very little friction.

1419
01:18:00.590 --> 01:18:03.433
But for high viscosity, there's a lot of resistance to that movement.

1420
01:18:03.713 --> 01:18:07.336
And therein lies the clue in how we can mathematically describe this.

1421
01:18:07.937 --> 01:18:11.460
Because we need to have a mathematical language to describe what is viscosity.

1422
01:18:12.000 --> 01:18:14.242
So let's go back to our traffic analogy.

1423
01:18:14.343 --> 01:18:14.643
Okay.

1424
01:18:14.723 --> 01:18:14.963
Okay?

1425
01:18:15.404 --> 01:18:17.726
Let's consider a multi-laned highway.

1426
01:18:18.599 --> 01:18:18.959
Like this.

1427
01:18:18.999 --> 01:18:22.783
So we've got a shoulder that's zero miles per hour, obviously, because you're on the shoulder.

1428
01:18:23.103 --> 01:18:24.524
You better be stationary.

1429
01:18:24.704 --> 01:18:29.588
You don't want to be one of those guys that tries to cut in line and then the cops catch him.

1430
01:18:30.049 --> 01:18:31.730
So let's look at what exchange of

1431
01:18:34.963 --> 01:18:51.358
traffic from one lane to another will do because in normal traffic obviously cars are going to change lanes yes so i've color-coded these cars based on their lane and based on their velocity sorry and what i'm realizing earlier you specifically showed us only one lane examples yes

1432
01:18:52.456 --> 01:18:54.658
Which was fine at the time.

1433
01:18:54.718 --> 01:18:55.199
At the time.

1434
01:18:55.639 --> 01:18:57.100
But now we've got multiple lanes.

1435
01:18:57.180 --> 01:18:59.523
And now the flow is going to crisscross.

1436
01:18:59.763 --> 01:19:03.306
In between, which is how real liquids... How real liquids do it, right?

1437
01:19:03.686 --> 01:19:08.191
Now, what I've done is, in the initial, let's say it's a four-laned highway...

1438
01:19:10.072 --> 01:19:12.893
The lane that's fastest is going at like 80 miles an hour.

1439
01:19:13.793 --> 01:19:16.173
And the lane that's slowest is going at 20 miles an hour.

1440
01:19:16.233 --> 01:19:19.954
So it goes 80, then 60, then 40, then 20, and then you've got the shoulder.

1441
01:19:19.994 --> 01:19:23.195
Four lanes, and it's in increments of 20 miles per hour.

1442
01:19:23.315 --> 01:19:25.296
And then things start shifting around.

1443
01:19:25.956 --> 01:19:26.216
Okay?

1444
01:19:27.636 --> 01:19:31.177
You actually wouldn't notice a change in the flow.

1445
01:19:32.138 --> 01:19:36.881
Because let's take, for example, the lane that's moving at 60 miles an hour.

1446
01:19:36.981 --> 01:19:40.944
It's going to inherit some faster cars from the 80 miles per hour lane.

1447
01:19:41.404 --> 01:19:44.566
And it's going to inherit some slower cars from the 40 miles per hour lane.

1448
01:19:44.887 --> 01:19:46.368
And the things are going to average out.

1449
01:19:47.556 --> 01:19:47.816
Right?

1450
01:19:48.757 --> 01:19:59.443
And so if I were to look at a graph, and that's on the bottom there, if we look at a graph of lane on the x-axis and the speed on the y-axis, it's a line because it goes from 0, 20, 40, 60, 80.

1451
01:19:59.924 --> 01:20:01.745
And later on, it's still going to be a line.

1452
01:20:02.558 --> 01:20:07.162
There's been no change, meaning that the velocity field has not changed, right?

1453
01:20:07.302 --> 01:20:09.804
It's still big up top, small down below.

1454
01:20:10.144 --> 01:20:14.448
And the velocity field has not changed from one time to another, meaning there's no force here.

1455
01:20:14.468 --> 01:20:15.028
Yeah, yeah, yeah.

1456
01:20:15.228 --> 01:20:15.729
Okay?

1457
01:20:15.829 --> 01:20:17.850
Even though there's some viscosity, there's no force.

1458
01:20:18.491 --> 01:20:18.931
Right, okay.

1459
01:20:19.051 --> 01:20:19.251
Yes.

1460
01:20:19.271 --> 01:20:19.512
Right?

1461
01:20:19.672 --> 01:20:19.872
Yes.

1462
01:20:20.072 --> 01:20:22.994
There's still like lane changing happening, but there's no force.

1463
01:20:24.035 --> 01:20:26.898
There's no change in the flow of traffic.

1464
01:20:27.878 --> 01:20:29.800
Now let's consider a different type of profile.

1465
01:20:31.690 --> 01:20:34.473
A type of profile where the shoulder is again zero.

1466
01:20:35.314 --> 01:20:38.697
And I've got the two slowest lanes, both at 20 miles per hour.

1467
01:20:39.338 --> 01:20:41.119
And then I've got a lane at 30 miles per hour.

1468
01:20:41.159 --> 01:20:43.121
But the faster lane is still going at 80.

1469
01:20:43.422 --> 01:20:43.922
Imagine that.

1470
01:20:44.202 --> 01:20:47.085
Like the carpool lane is still going at 80.

1471
01:20:47.386 --> 01:20:49.508
And then the rest is like still very, very slow.

1472
01:20:49.768 --> 01:20:51.350
And the carpool lane has like a barrier.

1473
01:20:52.207 --> 01:20:53.127
So there's nothing happening.

1474
01:20:53.187 --> 01:20:54.728
And then all of a sudden the barrier gets lifted.

1475
01:20:55.368 --> 01:20:56.028
What's going to happen?

1476
01:20:56.309 --> 01:20:58.129
Well, the carpool lane is going to slow down a lot.

1477
01:20:58.850 --> 01:21:06.493
But also the other lanes are going to speed up because stuff from the carpool lane is going to come in and sort of push that lane forward.

1478
01:21:06.793 --> 01:21:08.713
This is a bumper car scenario, okay?

1479
01:21:08.953 --> 01:21:09.494
Not real.

1480
01:21:10.674 --> 01:21:12.855
But you get what I'm saying?

1481
01:21:13.035 --> 01:21:15.116
Like before we had this...

1482
01:21:16.128 --> 01:21:23.912
Curvy profile, which we see on the lower left, where, again, we don't have a line because it's slow, slow, slow, then really fast.

1483
01:21:23.932 --> 01:21:24.152
Yes.

1484
01:21:24.932 --> 01:21:27.934
And with subsequent time, it's going to level out.

1485
01:21:27.954 --> 01:21:28.814
It's going to level out.

1486
01:21:29.195 --> 01:21:30.535
Because of the diffusion of cars.

1487
01:21:31.176 --> 01:21:33.097
Here, there is a force, right?

1488
01:21:33.137 --> 01:21:35.718
Because I've changed the velocity profile.

1489
01:21:36.058 --> 01:21:40.963
If I change the velocity profile, that's an acceleration, which means there's a force lurking in here.

1490
01:21:42.044 --> 01:21:42.224
Right?

1491
01:21:42.264 --> 01:21:45.167
There's a viscous force that is lurking in here.

1492
01:21:45.667 --> 01:21:56.157
And this is what's interesting is, just going back to our previous example, because basically the system was evenly distributed in the initial state...

1493
01:21:57.358 --> 01:22:01.599
based on how the variables would end up mixing, you had no net effect, no force.

1494
01:22:01.739 --> 01:22:02.199
No force.

1495
01:22:02.980 --> 01:22:09.641
In our second example here, it was sort of a not evenly distributed system.

1496
01:22:09.861 --> 01:22:13.282
And so when you look over time, and then there's the mixing.

1497
01:22:13.322 --> 01:22:14.003
Yeah, there's a mixing.

1498
01:22:14.023 --> 01:22:14.923
There's going to be a change.

1499
01:22:14.943 --> 01:22:15.723
There's some divergence.

1500
01:22:15.803 --> 01:22:16.183
Not divergence.

1501
01:22:16.203 --> 01:22:17.163
Let me not use the word divergence.

1502
01:22:17.183 --> 01:22:17.924
There's some change.

1503
01:22:17.944 --> 01:22:19.364
There's a change over time.

1504
01:22:19.384 --> 01:22:19.944
Over time.

1505
01:22:20.004 --> 01:22:21.145
Which means that there is a force.

1506
01:22:21.485 --> 01:22:26.406
And notice, in the first example, there was a straight line, which means I had a positive derivative

1507
01:22:26.986 --> 01:22:27.207
Right?

1508
01:22:27.267 --> 01:22:31.191
Like, the slope is positive, but the second derivative is zero.

1509
01:22:31.231 --> 01:22:32.272
There's no curvature.

1510
01:22:32.932 --> 01:22:35.315
In the second version, there's a curvature, right?

1511
01:22:35.355 --> 01:22:43.783
Which is why when I inherit the fast stuff and I inherit the slow stuff, the fast stuff is way faster than the slow stuff, which is why I'm speeding up.

1512
01:22:44.584 --> 01:22:48.528
The curvature is the origin of the force.

1513
01:22:49.532 --> 01:22:50.612
That's the key.

1514
01:22:50.692 --> 01:22:54.254
And so that is the form that the viscosity is going to take mathematically.

1515
01:22:54.274 --> 01:22:58.495
And if we go in the next slide, we'll actually see what the viscosity term looks like.

1516
01:22:59.035 --> 01:23:01.656
It looks like del squared.

1517
01:23:01.696 --> 01:23:04.957
Del squared means a second derivative in some sense with respect to space.

1518
01:23:05.797 --> 01:23:07.018
And there's your viscosity.

1519
01:23:07.058 --> 01:23:10.839
The mu, the Greek letter there, that's telling you how viscous it is.

1520
01:23:10.879 --> 01:23:13.219
Like honey would have a very high value.

1521
01:23:13.400 --> 01:23:14.840
Water would have a very low value.

1522
01:23:16.205 --> 01:23:17.548
And that's the force.

1523
01:23:18.029 --> 01:23:22.300
The force is, again, has to do with the velocity field itself.

1524
01:23:23.171 --> 01:23:23.991
This is always key.

1525
01:23:24.031 --> 01:23:31.013
We want to describe everything, except for maybe the external force, as a property of the velocity field itself.

1526
01:23:31.033 --> 01:23:41.795
Of whatever the system, as I use in a colloquial sense, as a whole, is a material idea here, which is why we have, these are all derivatives, which we're looking at the system level.

1527
01:23:41.815 --> 01:23:42.515
Yes, exactly.

1528
01:23:42.575 --> 01:23:45.916
And it kind of reminds me of when we were talking about the springs, right?

1529
01:23:46.236 --> 01:23:48.316
The force on the spring had to do with the position.

1530
01:23:48.916 --> 01:23:52.197
In this case, the force on the spring has to do with the velocity field itself.

1531
01:23:52.217 --> 01:23:52.657
Mm-hmm.

1532
01:23:53.738 --> 01:24:01.967
And on the right hand side, just for those who are engineers and physicists, I want to show you a similarity between the viscosity term and the

1533
01:24:03.016 --> 01:24:12.365
heat equation, where if you have the heat equation and if you have a temperature profile, as time moves forward, everything gets smoothed out because of thermal diffusion.

1534
01:24:12.966 --> 01:24:14.668
The same thing is happening with viscosity.

1535
01:24:14.708 --> 01:24:18.612
Viscosity is smoothing out the velocity differences, right?

1536
01:24:18.652 --> 01:24:20.834
And that's what we saw was happening with the traffic analogy.

1537
01:24:21.194 --> 01:24:25.739
There was a lot of high speed here and low speed here, and viscosity was smoothing it out.

1538
01:24:26.279 --> 01:24:54.153
and making the flow a bit smoother right i'm starting to see the connection between the words incompressible and smoothness that i keep seeing as it relates to the whole yeah navier stokes piece yeah we're we're starting to see that so the viscous term becomes a smoothing operator in some sense right um and now which feels very natural as a property of trying to measure a liquid because in this context it's it's

1539
01:24:54.633 --> 01:24:58.815
It's sort of this contiguous thingy when you look at it from the vector field perspective.

1540
01:24:59.135 --> 01:25:00.335
Yeah, exactly.

1541
01:25:00.515 --> 01:25:05.277
And we don't want that contiguous thingy to have like little pockets of really high fluid.

1542
01:25:05.537 --> 01:25:06.918
I'm sorry, really high velocity.

1543
01:25:07.218 --> 01:25:11.420
Because if there's a little pocket of really high velocity, well, that velocity is going to spread out.

1544
01:25:11.540 --> 01:25:11.720
Right.

1545
01:25:11.900 --> 01:25:12.080
Right?

1546
01:25:12.160 --> 01:25:13.400
And that's what viscosity does.

1547
01:25:13.461 --> 01:25:13.661
Right.

1548
01:25:14.101 --> 01:25:14.321
Right?

1549
01:25:14.741 --> 01:25:19.323
That's why in honey, it's really hard to have a small pocket of really high speed honey.

1550
01:25:19.343 --> 01:25:19.683
Yeah.

1551
01:25:20.607 --> 01:25:25.890
On the other hand, for water, it's maybe a little bit easier to have a pocket of high-speed water.

1552
01:25:26.230 --> 01:25:28.711
But then it just depends on what time... Yeah.

1553
01:25:28.871 --> 01:25:29.091
Anyway.

1554
01:25:29.111 --> 01:25:29.311
Yeah.

1555
01:25:29.351 --> 01:25:29.531
Yeah.

1556
01:25:30.172 --> 01:25:30.592
Exactly.

1557
01:25:30.612 --> 01:25:36.355
And so now we can solve D'Alembert's paradox because now we have viscosity in mind.

1558
01:25:36.635 --> 01:25:39.256
So we can apply this thing called the no-slip condition.

1559
01:25:39.496 --> 01:25:48.061
This is something that we've looked at with hypersonics and we've looked at in the FIFA episode, where if you've got an object that is moving through a fluid...

1560
01:25:49.381 --> 01:25:53.582
very close to the object, the fluid is not going to be moving at all.

1561
01:25:53.602 --> 01:25:54.903
It's going to be stuck to the object.

1562
01:25:55.363 --> 01:26:02.164
And there's going to be a boundary layer where it's going to transition between being stuck to the object and moving with the rest of the fluid.

1563
01:26:03.565 --> 01:26:09.286
And that boundary layer causes the drag because the object is dragging the fluid with it.

1564
01:26:10.026 --> 01:26:10.266
Right?

1565
01:26:11.387 --> 01:26:16.548
And this is how we solve D'Alembert's paradox because now if we look at, for example, a golf ball,

1566
01:26:17.373 --> 01:26:18.836
that is moving in air.

1567
01:26:19.478 --> 01:26:21.782
The fluid is sticking to the golf ball.

1568
01:26:22.764 --> 01:26:23.245
And so...

1569
01:26:24.162 --> 01:26:28.065
It's creating a wake behind it because the golf ball is moving through the fluid.

1570
01:26:28.826 --> 01:26:37.173
And as the fluid sticks to it, the part that's right above the golf ball is going to be moving with that sticky layer, but it's also going to be moving kind of with that.

1571
01:26:37.513 --> 01:26:41.016
And so you get all of these dynamics happening because of viscosity.

1572
01:26:41.476 --> 01:26:48.082
There's this transitionary space between the surface of the object that is moving through the fluid and then the rest of space.

1573
01:26:48.702 --> 01:27:01.525
And in proximity to the surface of the object, there is a transition where the fluid around the object is going to move on some gradient from sticking to the object to being a part of the larger system.

1574
01:27:01.645 --> 01:27:02.105
Exactly.

1575
01:27:02.205 --> 01:27:02.965
Yeah, exactly.

1576
01:27:03.225 --> 01:27:04.345
And so that's how we get drag.

1577
01:27:04.805 --> 01:27:05.886
And so this is the resolution.

1578
01:27:05.986 --> 01:27:09.526
And this is the contribution that Navier and Stokes were trying to capture.

1579
01:27:11.007 --> 01:27:12.867
Claude-Louis Navier said,

1580
01:27:13.538 --> 01:27:17.962
in 1822, he presented an equation on viscous fluid motion where he added that viscous term.

1581
01:27:18.363 --> 01:27:20.845
But he was an engineer as much as a mathematician.

1582
01:27:21.306 --> 01:27:26.771
And he tried to explain the internal friction by imagining molecular interactions inside a fluid.

1583
01:27:26.811 --> 01:27:27.351
This was 1822.

1584
01:27:27.411 --> 01:27:31.835
So everyone thought he was crazy because there was no such thing as atoms back then.

1585
01:27:32.076 --> 01:27:35.859
I mean, there was this Greek concept of an atom by democratized or something.

1586
01:27:36.139 --> 01:27:36.540
But, you know...

1587
01:27:37.120 --> 01:27:38.341
No one took that seriously.

1588
01:27:38.801 --> 01:27:46.124
And it's one of these rare occasions where a good outcome came from a derivation that's based on a questionable premise.

1589
01:27:47.065 --> 01:27:48.726
Now we know it's not questionable, obviously.

1590
01:27:49.026 --> 01:27:51.607
And over the following decades, we had Cauchy, Poisson…

1591
01:27:52.427 --> 01:27:57.952
And others developed the continuum theory of stress and how stress affects bodies.

1592
01:27:58.732 --> 01:28:10.422
And in 1845, Stokes, George Gabriel Stokes, derived the viscous equations from a completely different direction from this idea of stress and how you can deform objects.

1593
01:28:10.963 --> 01:28:13.245
Turns out it's the same form as Navier.

1594
01:28:13.605 --> 01:28:15.527
And so now that's why we're called the Navier-Stokes equation.

1595
01:28:16.178 --> 01:28:16.699
equations.

1596
01:28:17.059 --> 01:28:18.420
They never collaborated, by the way.

1597
01:28:18.440 --> 01:28:26.367
Stokes was a teenager when Navier died, but their names are joined because, you know, different physical arguments, but they arrived at these two equations.

1598
01:28:26.487 --> 01:28:32.213
And now these are the two equations that have been such a headache for mathematicians and physicists all these years.

1599
01:28:32.893 --> 01:28:45.324
And so as I look at now these two equations, the one at the bottom, which was where we started, is this idea that the divergence always has to be zero for an incompressible fluid.

1600
01:28:45.344 --> 01:28:45.504
Yeah.

1601
01:28:45.624 --> 01:28:48.166
You can't make fluid and you can't destroy fluid.

1602
01:28:48.186 --> 01:28:55.772
You can't have a fluid emanate from a point in all directions, and you can't have a fluid converge on a point from all directions.

1603
01:28:56.233 --> 01:28:58.074
That's what the bottom one is effectively saying.

1604
01:28:58.134 --> 01:28:58.235
Yep.

1605
01:28:58.255 --> 01:28:58.355
Yep.

1606
01:28:58.896 --> 01:29:11.161
And then now the top one, which was the more complex one, is we're trying to replicate force on the right equals mass times acceleration on the left.

1607
01:29:11.221 --> 01:29:11.581
Yes.

1608
01:29:11.661 --> 01:29:12.342
In this construction.

1609
01:29:12.382 --> 01:29:12.762
Exactly.

1610
01:29:12.842 --> 01:29:15.523
Which was Newton's second law of motion.

1611
01:29:16.024 --> 01:29:18.485
We're trying to apply it to incompressible fluids.

1612
01:29:18.505 --> 01:29:19.606
Yes.

1613
01:29:19.946 --> 01:29:28.351
And in doing so, we constructed the left side of the equation first where we have our mass represented by our volume.

1614
01:29:28.531 --> 01:29:30.492
Yeah, that's the density.

1615
01:29:30.572 --> 01:29:31.433
The density, excuse me.

1616
01:29:32.293 --> 01:29:36.676
And then we then multiply that by everything in the larger brackets, which is our acceleration.

1617
01:29:36.696 --> 01:29:37.116
Yeah.

1618
01:29:37.336 --> 01:29:43.921
which we have the first term being the time derivative of the velocity field.

1619
01:29:43.941 --> 01:29:46.923
Yeah, that's like rain coming in and changing traffic patterns.

1620
01:29:46.943 --> 01:29:47.324
Patterns.

1621
01:29:48.725 --> 01:29:57.311
But we also have to add into that now the convective term, which is how an individual car within that vector field works.

1622
01:29:57.831 --> 01:30:03.136
is changing its velocity with distance over time.

1623
01:30:03.897 --> 01:30:04.898
And so that's the left side.

1624
01:30:04.938 --> 01:30:05.158
Yeah.

1625
01:30:05.438 --> 01:30:08.461
And that gives us our mass times acceleration.

1626
01:30:08.501 --> 01:30:08.721
Yes.

1627
01:30:09.742 --> 01:30:10.422
Loosely speaking.

1628
01:30:10.642 --> 01:30:12.404
Maybe there's a few details not quite right.

1629
01:30:12.424 --> 01:30:13.385
And then on our right…

1630
01:30:14.837 --> 01:30:21.880
So there's sort of three terms that have complexity on our mass times acceleration on the left.

1631
01:30:22.060 --> 01:30:23.881
It's really two terms, I would say, right?

1632
01:30:23.921 --> 01:30:26.522
Because it's – Sorry, fair.

1633
01:30:26.802 --> 01:30:29.243
Within acceleration, it has two component parts.

1634
01:30:29.423 --> 01:30:30.843
Well, I'm nitpicking here.

1635
01:30:30.883 --> 01:30:34.225
But it's really two terms because it's density multiplied by du dt.

1636
01:30:34.585 --> 01:30:34.845
Okay.

1637
01:30:34.885 --> 01:30:36.565
And then it's density multiplied by that second term.

1638
01:30:36.585 --> 01:30:36.866
Totally fair.

1639
01:30:36.886 --> 01:30:37.406
You know what I mean?

1640
01:30:37.426 --> 01:30:37.686
Totally fair.

1641
01:30:37.706 --> 01:30:39.387
Like density is not its own thing.

1642
01:30:39.407 --> 01:30:42.168
That's a fair – No, that's an important distinction.

1643
01:30:42.268 --> 01:30:43.088
And that's helpful.

1644
01:30:43.993 --> 01:30:51.343
And then on our right side, where we're trying to define force, we started with the obvious thing, which was pressure.

1645
01:30:51.424 --> 01:30:51.644
Yeah.

1646
01:30:51.884 --> 01:30:54.107
And it's negative because high pressure always moves to low pressure.

1647
01:30:54.167 --> 01:30:54.368
Mm-hmm.

1648
01:30:55.753 --> 01:31:03.217
Then we moved to – we added our F, which is some general external force.

1649
01:31:03.338 --> 01:31:05.679
Yeah, stirring, gravity, whatever you wanted to call it.

1650
01:31:06.139 --> 01:31:08.240
And that's where Euler ended up.

1651
01:31:08.320 --> 01:31:08.561
Yeah.

1652
01:31:09.001 --> 01:31:09.641
And he stopped.

1653
01:31:09.721 --> 01:31:10.282
And he stopped.

1654
01:31:11.162 --> 01:31:14.724
And we were missing this concept of drag or resistance.

1655
01:31:15.525 --> 01:31:15.665
And –

1656
01:31:16.385 --> 01:31:31.928
Ultimately, both Navier and Stokes, from different directions, ended up on being able to define viscosity as the missing piece, which is, again, talking about there's a boundary layer between the object in the system and then the rest of the medium external environment.

1657
01:31:32.288 --> 01:31:34.688
And there's some gradient delta there of change.

1658
01:31:34.708 --> 01:31:39.569
And so we need to account for the viscosity because that applies a force as we look at all of this.

1659
01:31:39.869 --> 01:31:43.050
Is that now like an accurate description of what we're looking at?

1660
01:31:43.070 --> 01:31:44.731
That is the Navier Stokes equation.

1661
01:31:44.771 --> 01:31:48.514
And we have just built it up from first principles.

1662
01:31:48.674 --> 01:31:56.200
As someone who did not do any type of mathematics, it's a testament to your way of taking the story from start to finish.

1663
01:31:56.220 --> 01:31:56.380
Yeah.

1664
01:31:56.500 --> 01:31:57.641
And it kind of makes sense though, right?

1665
01:31:57.661 --> 01:31:58.682
It makes total sense.

1666
01:31:58.842 --> 01:32:01.004
When you understand the underlying systems.

1667
01:32:01.024 --> 01:32:01.944
Yeah.

1668
01:32:02.445 --> 01:32:06.628
I think the hardest part really is getting familiar with the idea of a velocity field.

1669
01:32:06.828 --> 01:32:07.008
Yeah.

1670
01:32:07.028 --> 01:32:07.128
Yeah.

1671
01:32:07.148 --> 01:32:07.389
You know?

1672
01:32:07.549 --> 01:32:07.649
Yeah.

1673
01:32:07.689 --> 01:32:11.812
Because once you have that and you've internalized, oh, it's like a traffic pattern.

1674
01:32:12.506 --> 01:32:16.707
Then all of the other stuff kind of falls together because you can kind of figure it out.

1675
01:32:17.087 --> 01:32:20.429
That was the Eulerian perspective was the velocity field.

1676
01:32:20.449 --> 01:32:21.629
That's the way we described it earlier.

1677
01:32:21.769 --> 01:32:22.029
Yeah.

1678
01:32:22.209 --> 01:32:22.709
Very nice.

1679
01:32:22.829 --> 01:32:24.590
D'Alembert, dude.

1680
01:32:24.650 --> 01:32:25.430
D'Alembert.

1681
01:32:25.750 --> 01:32:26.070
Yeah.

1682
01:32:26.931 --> 01:32:31.352
And so now we can start asking, at this point, we've got the Navier-Stokes equations.

1683
01:32:31.552 --> 01:32:31.772
Yes.

1684
01:32:31.792 --> 01:32:32.933
We know what every term means.

1685
01:32:33.153 --> 01:32:33.373
Yes.

1686
01:32:33.593 --> 01:32:33.873
So...

1687
01:32:35.007 --> 01:32:37.808
Now we can start tackling how to solve them.

1688
01:32:38.729 --> 01:32:38.969
Right.

1689
01:32:39.649 --> 01:32:44.652
And because part of what we also define through all of this is this is an extremely complex problem.

1690
01:32:45.247 --> 01:33:07.352
problem yes for all of the nuances of each of the underlying terms which also have their own sub uh issues some subcategory issues exactly and so we've got an equation of motion now can we solve it like can we can we find a general like i give you a velocity field now you tell me what the velocity field is later right what is the flow going to look like later and

1691
01:33:07.692 --> 01:33:17.415
And to just come back to the beginning of the episode, the reason we want to do this is because with Newton's equations that we were talking about, part of it's like then you can use it to make predictions.

1692
01:33:17.595 --> 01:33:18.035
Exactly.

1693
01:33:18.175 --> 01:33:18.375
Yeah.

1694
01:33:18.635 --> 01:33:21.076
And that's how we do the space stuff and blah, blah, blah.

1695
01:33:21.096 --> 01:33:26.177
So what we're trying to do is take this equation of motion so that we can – like what is – who cares?

1696
01:33:26.297 --> 01:33:26.497
Yeah.

1697
01:33:26.557 --> 01:33:28.558
And the answer is so we can make predictions.

1698
01:33:28.578 --> 01:33:34.519
We can give you one input, the velocity field, and then you can give me the output consistently that's true every time.

1699
01:33:34.579 --> 01:33:34.739
Yeah.

1700
01:33:34.759 --> 01:33:35.240
That'd be dope.

1701
01:33:35.440 --> 01:33:35.920
That'd be great.

1702
01:33:36.120 --> 01:33:36.320
Yeah.

1703
01:33:37.060 --> 01:33:42.683
Okay, so we have a set of differential equations, and we'd like to have solutions to those differential equations.

1704
01:33:43.183 --> 01:33:44.123
Now, what do we mean by that?

1705
01:33:44.163 --> 01:33:50.926
Like, in a normal algebra problem, if I write something like x plus 2 equals 5, I know what x is.

1706
01:33:51.026 --> 01:33:51.767
x is 3, right?

1707
01:33:51.807 --> 01:33:52.787
I can, like, do the thing.

1708
01:33:53.147 --> 01:33:55.588
In a differential equation, what we're asking is something larger.

1709
01:33:55.668 --> 01:33:59.150
What is the function that solves the differential equation?

1710
01:33:59.590 --> 01:34:01.951
In our case, it would be what is the velocity field that solves the differential equation?

1711
01:34:02.031 --> 01:34:03.972
That solves this differential equation.

1712
01:34:04.272 --> 01:34:05.693
Let's take a simple example.

1713
01:34:06.393 --> 01:34:13.076
Let's take, for example, the derivative of y in terms of t is equal to some r times y.

1714
01:34:13.436 --> 01:34:15.597
Meaning, I've got some quantity.

1715
01:34:16.057 --> 01:34:22.580
The change of that quantity with respect to time is proportional to the quantity itself multiplied by, let's say, some rate.

1716
01:34:23.757 --> 01:34:25.539
That is an exponential, right?

1717
01:34:25.959 --> 01:34:32.424
The solution to that is e to the power of rt, and this is how people get rich.

1718
01:34:32.585 --> 01:34:34.266
This is compound interest, right?

1719
01:34:34.286 --> 01:34:38.169
You start with some amount of money, and then the rate tells you the interest over time.

1720
01:34:38.610 --> 01:34:41.232
And the more money you have, the faster your money grows.

1721
01:34:41.753 --> 01:34:46.677
And this is a very simple differential equation that we can solve very accurately.

1722
01:34:47.097 --> 01:34:47.277
Right.

1723
01:34:47.698 --> 01:34:49.979
This is a toy equation and the answer is very nice.

1724
01:34:51.521 --> 01:34:54.863
Navier Stokes is less accommodating.

1725
01:34:54.883 --> 01:34:55.063
Okay.

1726
01:34:55.524 --> 01:35:01.168
The unknown is not how much money you have, like a single one-dimensional variable.

1727
01:35:01.629 --> 01:35:03.650
It's a three-dimensional velocity field.

1728
01:35:03.730 --> 01:35:03.971
Yeah.

1729
01:35:04.011 --> 01:35:04.211
Right.

1730
01:35:04.551 --> 01:35:04.791
Yeah.

1731
01:35:04.811 --> 01:35:06.393
So it's a number.

1732
01:35:06.413 --> 01:35:06.513
Yeah.

1733
01:35:07.627 --> 01:35:12.208
at every point in 3D space, but that number has a direction, right?

1734
01:35:12.308 --> 01:35:15.809
It's a length, and it's got a direction that's also three-dimensional.

1735
01:35:16.489 --> 01:35:20.189
So it's quite massive, like this equation.

1736
01:35:21.430 --> 01:35:31.012
And it's so funny, because now that I understand that, it makes my brain hurt trying to say, we want to solve for this now, given that level of complexity, which is now very clear what that level of complexity is.

1737
01:35:31.032 --> 01:35:31.512
Exactly.

1738
01:35:31.612 --> 01:35:32.032
Like the...

1739
01:35:33.132 --> 01:35:35.034
Fluid is interacting with itself and everything.

1740
01:35:35.314 --> 01:35:39.779
So this velocity field has to solve for that entire big thing, right?

1741
01:35:40.019 --> 01:35:42.221
We've got to find all of the arrows everywhere.

1742
01:35:42.501 --> 01:35:42.721
Right.

1743
01:35:43.282 --> 01:35:43.762
In some sense.

1744
01:35:43.802 --> 01:35:44.043
Yes.

1745
01:35:44.203 --> 01:35:44.403
Right?

1746
01:35:44.463 --> 01:35:45.204
For all time.

1747
01:35:45.224 --> 01:35:47.106
Because the arrows might be changing over time.

1748
01:35:47.146 --> 01:35:47.566
Over time.

1749
01:35:47.826 --> 01:35:49.848
And we need to have an expression for all of that.

1750
01:35:50.189 --> 01:35:56.235
Now, there are certain special situations in which the Navier-Stokes is actually exactly solvable.

1751
01:35:56.975 --> 01:36:00.977
Okay, it's not true that the Navier-Stokes is just unsolvable.

1752
01:36:01.057 --> 01:36:02.497
You can't solve this equation.

1753
01:36:02.918 --> 01:36:09.740
There are situations, there are certain boundary conditions where you can exactly solve the Navier-Stokes equations.

1754
01:36:09.820 --> 01:36:10.681
Let's go through one of them.

1755
01:36:10.721 --> 01:36:14.602
It's called plain co-et, co-et, co-et-ay flow.

1756
01:36:15.063 --> 01:36:16.223
I never know how to pronounce it.

1757
01:36:16.263 --> 01:36:17.404
I think it might be co-et.

1758
01:36:17.704 --> 01:36:18.784
Is it co-et-ay?

1759
01:36:18.964 --> 01:36:19.985
Let us know in the comments.

1760
01:36:20.045 --> 01:36:20.145
Yeah.

1761
01:36:20.165 --> 01:36:20.305
Yeah.

1762
01:36:21.064 --> 01:36:22.085
In any case, it looks French.

1763
01:36:22.345 --> 01:36:25.728
Yeah, it's probably French, which is why I think it's cohete.

1764
01:36:26.769 --> 01:36:27.429
Here's what's happening.

1765
01:36:27.449 --> 01:36:29.271
You've got two plates, okay?

1766
01:36:29.491 --> 01:36:31.593
Those are the Navier-Stokes equations that we're trying to solve.

1767
01:36:31.813 --> 01:36:32.013
Yes.

1768
01:36:32.353 --> 01:36:33.134
I've got two plates.

1769
01:36:33.354 --> 01:36:37.598
One plate, let's say, is the table, completely stationary.

1770
01:36:38.018 --> 01:36:43.243
Another infinite plate is moving in the x direction, in just one direction at a certain velocity.

1771
01:36:43.563 --> 01:36:45.985
And in between two plates, you've got water.

1772
01:36:46.105 --> 01:36:47.026
You've got some kind of fluid.

1773
01:36:47.588 --> 01:36:47.808
Okay?

1774
01:36:48.309 --> 01:36:52.192
What is the velocity profile of the water in between these two plates?

1775
01:36:52.892 --> 01:36:53.713
It's very simple, right?

1776
01:36:54.113 --> 01:37:01.039
There's a stationary plate, there's a plate on top that is moving at a fixed velocity in one direction, and there's fluid in the middle.

1777
01:37:01.439 --> 01:37:04.482
So what is the velocity field going through?

1778
01:37:04.742 --> 01:37:06.364
Now we know from the no-slip condition...

1779
01:37:07.689 --> 01:37:11.231
The fluid that's right next to the table that's stationary is going to be stationary.

1780
01:37:11.571 --> 01:37:14.913
And the fluid that's on the plate that's moving is going to be moving with the plate.

1781
01:37:15.213 --> 01:37:16.093
What happens in the middle?

1782
01:37:16.774 --> 01:37:17.894
It's actually pretty simple.

1783
01:37:18.435 --> 01:37:20.556
You solve for the Navier-Stokes equations.

1784
01:37:21.760 --> 01:37:24.222
And you get this profile, right?

1785
01:37:25.083 --> 01:37:28.625
Near the bottom, where it's stationary, the velocity is zero.

1786
01:37:29.206 --> 01:37:33.969
And as you approach the plate in the top, the velocity gradually increases linearly.

1787
01:37:34.370 --> 01:37:36.531
And so it's just a function of the height, right?

1788
01:37:36.771 --> 01:37:40.975
The higher you are, the closer you are to the top plate, the closer you are to the velocity.

1789
01:37:41.035 --> 01:37:42.536
It's a linear relationship.

1790
01:37:42.796 --> 01:37:42.936
Boom.

1791
01:37:42.956 --> 01:37:43.617
That makes total sense.

1792
01:37:43.637 --> 01:37:44.697
Exactly solvable.

1793
01:37:44.737 --> 01:37:45.918
That makes total sense.

1794
01:37:46.098 --> 01:37:56.004
And again, we've sort of created an exactly solvable, simple use case because, you know, by having the bottom plate be stationary.

1795
01:37:56.184 --> 01:37:56.385
Yeah.

1796
01:37:56.485 --> 01:38:00.127
Having the bottom, even if the bottom plate is actually moving in the other direction, it's totally fine.

1797
01:38:00.167 --> 01:38:03.149
As long as it's the same, like, you know, the same velocity.

1798
01:38:03.569 --> 01:38:05.130
And there's so much symmetry here.

1799
01:38:05.230 --> 01:38:06.832
They're only moving it on one plane.

1800
01:38:06.912 --> 01:38:07.353
Exactly.

1801
01:38:07.393 --> 01:38:09.195
It's only one coordinate that we have to worry about.

1802
01:38:09.456 --> 01:38:12.299
So a bunch of the derivatives over there just go to zero.

1803
01:38:12.620 --> 01:38:19.529
Like there's so much symmetry and geometry here that it murders all of the difficult terms of that equation.

1804
01:38:19.870 --> 01:38:22.273
And you've got this steady profile and you're good to go.

1805
01:38:22.673 --> 01:38:29.598
It's a recurring theme in fluid mechanics, which is if you've got a nice symmetric distribution of stuff, you can solve things.

1806
01:38:29.678 --> 01:38:29.978
Fine.

1807
01:38:30.178 --> 01:38:30.658
You're fine.

1808
01:38:30.838 --> 01:38:31.038
Yeah.

1809
01:38:31.078 --> 01:38:32.479
This is called co-wethe flow.

1810
01:38:32.780 --> 01:38:34.261
You can also do pipe flows.

1811
01:38:34.381 --> 01:38:42.626
There's a handful of vortices and jets where you can obtain beautiful closed-form solutions based on the symmetry of the situation.

1812
01:38:43.086 --> 01:38:55.069
And just to ask as a brief side question, part of the other part of why those are so solvable is because we have such a good understanding of geometry as like these other mathematical disciplines.

1813
01:38:55.370 --> 01:39:00.511
We can apply some of those learnings and concepts in a way that make these easy or it's not relevant at all.

1814
01:39:00.571 --> 01:39:01.251
Well, no, certainly.

1815
01:39:01.291 --> 01:39:06.454
I think, I mean, for that one, it's like pretty easy because you can just take derivatives and they go to zero and whatever, right?

1816
01:39:06.694 --> 01:39:13.076
But there's other ones, like there's certain vortices and jets where you can manipulate the symmetry to cancel out terms and things like that.

1817
01:39:13.377 --> 01:39:16.318
You can use our knowledge of symmetry

1818
01:39:16.958 --> 01:39:17.178
Right.

1819
01:39:17.278 --> 01:39:21.899
To make, like, transformations that make these equations much easier to handle.

1820
01:39:22.480 --> 01:39:28.081
The subtle note I'm just trying to make here is these things don't happen in a vacuum.

1821
01:39:28.401 --> 01:39:34.263
Like, our general mathematical understanding has applications in subtle, small ways all through.

1822
01:39:34.403 --> 01:39:34.923
All through, yeah.

1823
01:39:34.963 --> 01:39:35.343
All through.

1824
01:39:35.503 --> 01:39:36.184
Yeah, all through.

1825
01:39:36.484 --> 01:39:36.964
And so...

1826
01:39:38.384 --> 01:39:40.045
So that's for a very nice symmetric thing.

1827
01:39:40.505 --> 01:39:42.845
Formula One cars, not very symmetric.

1828
01:39:42.945 --> 01:39:45.426
I mean, there's some symmetry, I guess, left and right, right?

1829
01:39:45.466 --> 01:39:49.907
So sure, you can see that it's symmetric, like you can flip the profile, right?

1830
01:39:50.207 --> 01:39:52.288
But a Formula One car is insane.

1831
01:39:52.328 --> 01:39:54.188
The geometry is extremely complicated.

1832
01:39:54.488 --> 01:39:57.249
The wheels rotate, the ground moves relative to the car.

1833
01:39:57.529 --> 01:40:00.850
There's thin boundary layers on the bodywork entirely.

1834
01:40:01.250 --> 01:40:01.670
Air is...

1835
01:40:02.190 --> 01:40:15.626
accelerated beneath the floor the vortices are deliberately generated around these waves to cause downforce um the non-linear term does not vanish in the other one the non-linear term actually vanishes

1836
01:40:16.622 --> 01:40:18.083
just because of the geometry of the thing.

1837
01:40:18.424 --> 01:40:20.566
Here, this is the entire problem, right?

1838
01:40:20.606 --> 01:40:27.732
There's no known formula that you can write out for what a Formula 1 car, the velocity and pressure at every point around a Formula 1 car is going to use.

1839
01:40:28.012 --> 01:40:40.844
So instead, what engineers do, even in this visualization over here, what they're doing is using a computer to computationally solve the Navier-Stokes equation around this boundary.

1840
01:40:42.361 --> 01:40:44.662
And this is called computational fluid dynamics.

1841
01:40:45.062 --> 01:40:55.168
If there's any aerospace engineers or hydrodynamic engineers in the audience, leave one in the comments and tell us your favorite computational fluid dynamics program.

1842
01:40:55.468 --> 01:40:56.969
There's a bunch of programs out there.

1843
01:40:57.009 --> 01:40:58.911
This is one that I just picked up on GitHub.

1844
01:40:59.451 --> 01:41:07.616
And they always flaunt the number of cells that they can break down their space into.

1845
01:41:07.636 --> 01:41:07.736
Yeah.

1846
01:41:07.956 --> 01:41:11.440
Because, you know, imagine a Formula 1 car in a wind tunnel, right?

1847
01:41:11.480 --> 01:41:12.502
That's some, like, volume.

1848
01:41:13.383 --> 01:41:18.089
In order to solve these differential equations, you have to break up that volume into tiny little grid boxes.

1849
01:41:18.509 --> 01:41:21.873
And then you can solve the Navier-Stokes equations for each of the grid boxes.

1850
01:41:22.134 --> 01:41:23.635
And then you just can put them all together.

1851
01:41:23.655 --> 01:41:23.856
Yeah.

1852
01:41:24.116 --> 01:41:28.178
You can be like, oh, the velocity is here, but the velocity next to it is something.

1853
01:41:28.198 --> 01:41:30.279
The velocity to the right and left of it is something.

1854
01:41:30.519 --> 01:41:33.180
So I can calculate a derivative that way, right?

1855
01:41:33.220 --> 01:41:38.122
I can be like, oh, the difference divided by the grid spacing is my spatial derivative.

1856
01:41:38.142 --> 01:41:38.643
Yeah.

1857
01:41:38.823 --> 01:41:39.823
Right?

1858
01:41:39.923 --> 01:41:42.264
And I can do this computationally one at a time.

1859
01:41:42.504 --> 01:41:47.707
And the more cells I have, the better granularity and the more accuracy I have to the real thing.

1860
01:41:49.007 --> 01:41:52.209
This makes me think back to our hypersonics episode.

1861
01:41:52.598 --> 01:41:54.439
where we effectively discuss the same thing.

1862
01:41:54.599 --> 01:41:54.819
Yeah.

1863
01:41:55.599 --> 01:42:06.925
And then sort of in addition to all of the challenges we've talked about, when you start going from supersonic to hypersonic, things get even weirder.

1864
01:42:07.025 --> 01:42:07.265
Yeah.

1865
01:42:07.705 --> 01:42:10.967
And we sort of discussed why it gets weirder and how we try to –

1866
01:42:11.843 --> 01:42:15.889
deal with that with these approximations that we've sort of discussed here.

1867
01:42:15.909 --> 01:42:19.113
It's again, Navier Stokes is not unsolved in all cases.

1868
01:42:19.133 --> 01:42:20.415
We found clever ways.

1869
01:42:20.515 --> 01:42:20.795
Yeah.

1870
01:42:20.855 --> 01:42:21.436
To like approximate.

1871
01:42:21.476 --> 01:42:22.397
Approximate.

1872
01:42:23.455 --> 01:42:24.595
Yeah, the solution, right?

1873
01:42:24.615 --> 01:42:24.775
Right.

1874
01:42:24.916 --> 01:42:26.556
And for engineers, this is good enough.

1875
01:42:26.596 --> 01:42:26.796
Right.

1876
01:42:26.836 --> 01:42:28.857
The point, they're like, honestly, like, yeah.

1877
01:42:28.997 --> 01:42:30.057
Oh, but I didn't solve anything.

1878
01:42:30.077 --> 01:42:33.559
We've been modeling and having a stroke since before you were born.

1879
01:42:33.719 --> 01:42:39.761
Yeah, for the Ferrari engineers, it's not good enough because apparently they still can't do it.

1880
01:42:39.781 --> 01:42:41.782
Like, guys, figure it out.

1881
01:42:41.802 --> 01:42:44.162
You have Lewis Hamilton and Charles Leclerc.

1882
01:42:44.703 --> 01:42:46.183
Like, guys, come, please.

1883
01:42:46.443 --> 01:42:47.864
I've been waiting for so long.

1884
01:42:48.888 --> 01:42:51.311
Anyways, the result is not a closed-form solution, okay?

1885
01:42:51.391 --> 01:42:56.817
It's this enormous table of numbers that tells you what the velocity is at every given point.

1886
01:42:56.877 --> 01:42:59.300
And you can put any arbitrary shape into it.

1887
01:42:59.620 --> 01:43:02.383
This is like a meme now in aerodynamics for some reason.

1888
01:43:02.684 --> 01:43:05.567
Like, all these aerospace engineers, like, put a cow...

1889
01:43:06.067 --> 01:43:28.282
a cad model of a cow into their cfd simulations and they're like how aerodynamic is a cow i've seen so many versions of this meme for some reason i don't know why but a cow is a very non-aerodynamic thing that people are trying to figure out you know what is the lift on a cow as it goes through a wind tunnel in any case the idea is any arbitrary shape you can put it into a

1890
01:43:30.203 --> 01:43:37.392
And you can turn the simulation one after the other and computationally solve the Navier-Stokes equations.

1891
01:43:38.053 --> 01:43:40.616
For all practical purposes, that's totally fine.

1892
01:43:40.796 --> 01:43:42.318
It's an engineering solution.

1893
01:43:43.219 --> 01:43:45.361
It's not a physics understanding.

1894
01:43:45.401 --> 01:43:45.541
Yeah.

1895
01:43:45.561 --> 01:43:45.662
Yeah.

1896
01:43:46.378 --> 01:43:47.800
And or mathematical understanding.

1897
01:43:48.120 --> 01:43:51.644
And it's fine that it's an applied engineering solution.

1898
01:43:51.664 --> 01:43:54.066
It's a spectacular practical achievement.

1899
01:43:54.106 --> 01:43:56.789
I mean, this is how we get airplanes.

1900
01:43:56.889 --> 01:43:58.211
This is how we get fighter jets.

1901
01:43:58.251 --> 01:43:59.352
This is how we get turbines.

1902
01:43:59.692 --> 01:44:04.438
This is how we get wind farms for power.

1903
01:44:04.658 --> 01:44:06.079
But it's not the same thing as...

1904
01:44:06.878 --> 01:44:12.081
the exact equations, and what do those equations look like, right?

1905
01:44:12.121 --> 01:44:16.603
What does the solution to this thing look like for any arbitrary boundary condition?

1906
01:44:17.644 --> 01:44:19.484
A simulation has a finite grid, right?

1907
01:44:19.865 --> 01:44:22.246
It advances in finite time steps.

1908
01:44:22.706 --> 01:44:24.207
It rounds numbers at some point.

1909
01:44:25.467 --> 01:44:29.890
And so the problem is engineers are pretty happy.

1910
01:44:30.770 --> 01:44:38.635
Taking Navier-Stokes equations, putting it into a CAD, like putting CAD model in there, using computational fluid dynamics, they are good to go.

1911
01:44:39.095 --> 01:44:40.396
They'll tell you what the drag is.

1912
01:44:40.416 --> 01:44:44.659
They'll tell you what the lift is, what the downforce is on a Formula One car and so on and so forth.

1913
01:44:44.839 --> 01:44:47.360
And unless you're making hypersonics, it's fine.

1914
01:44:47.500 --> 01:44:47.941
It's fine.

1915
01:44:48.361 --> 01:44:57.907
Mathematicians, on the other hand, they're asking, well, what about like the existential problem of like I've got differential equations.

1916
01:44:58.287 --> 01:44:58.507
Yeah.

1917
01:44:59.393 --> 01:45:01.435
are the solutions always well-behaved?

1918
01:45:02.436 --> 01:45:02.716
Okay.

1919
01:45:03.096 --> 01:45:03.937
That's the question.

1920
01:45:04.658 --> 01:45:06.059
Here's what I mean by well-behaved.

1921
01:45:06.340 --> 01:45:10.564
Let me give you an example of why they shouldn't be well-behaved.

1922
01:45:10.964 --> 01:45:14.648
It really has to do with these two terms that are in complete...

1923
01:45:16.704 --> 01:45:18.366
competition to one another.

1924
01:45:18.706 --> 01:45:20.047
And let me see if I can get this right.

1925
01:45:20.067 --> 01:45:28.675
We're talking about the convective term on the left, which accounts for acceleration, and we're talking about the viscosity term on the right, which is a part of our force.

1926
01:45:28.755 --> 01:45:29.015
Yes.

1927
01:45:29.676 --> 01:45:32.078
And these two things do very different things.

1928
01:45:33.205 --> 01:45:36.147
qualitative transformations to our velocity field.

1929
01:45:36.627 --> 01:45:39.989
The viscosity term, as I told you, smooths things out, right?

1930
01:45:40.089 --> 01:45:47.333
If there's a tiny bit where there's a lot of velocity and it's surrounded by not that much velocity, then it's going to smooth out.

1931
01:45:47.353 --> 01:45:51.975
The velocity is going to leak through and you're going to smooth out the velocity function.

1932
01:45:53.096 --> 01:45:59.079
The convective term, on the other hand, what it does is transfer energy from large...

1933
01:46:00.693 --> 01:46:03.754
Systems to smaller and smaller systems.

1934
01:46:04.174 --> 01:46:12.856
So the convective term is making smaller and smaller structures, while the viscosity term is making those smaller structures smooth out.

1935
01:46:13.336 --> 01:46:17.838
There's a constant competition between these two terms in the equation.

1936
01:46:18.158 --> 01:46:25.399
Because depending on the size of the structure we're referring to, the viscosity will be different and the convective term will be different.

1937
01:46:25.760 --> 01:46:27.220
And it's different across every scale.

1938
01:46:27.320 --> 01:46:27.580
Yes.

1939
01:46:28.061 --> 01:46:30.423
And the question is, does one win out over the other?

1940
01:46:30.703 --> 01:46:34.686
Specifically, does the convective term ever win out over the viscosity term?

1941
01:46:35.006 --> 01:46:41.492
Can I get smaller and smaller structure to such an extent where I can have, theoretically, infinite velocity somewhere?

1942
01:46:41.772 --> 01:46:45.615
Because the convective term is just creating smaller and smaller structure.

1943
01:46:45.895 --> 01:46:51.039
Now, I want to be a bit more concrete here when I say the convective term is creating smaller and smaller structure.

1944
01:46:51.380 --> 01:46:54.783
Okay, so we're going to talk about this competition in a very...

1945
01:46:55.503 --> 01:46:58.186
real sense with something called the Taylor Green Vortex.

1946
01:46:58.546 --> 01:47:08.936
This is a standard test case that is used in computational fluid dynamics because what you can do is actually solve this thing analytically, and then you can solve it in a computer and compare the two.

1947
01:47:09.496 --> 01:47:15.442
And then that'll tell you how good your computer simulation is because you have a ground truth that's like a functional mathematical form.

1948
01:47:15.982 --> 01:47:18.325
Here's what the Taylor Green Vortex looks like.

1949
01:47:18.605 --> 01:47:19.826
This is the initial condition.

1950
01:47:19.846 --> 01:47:20.086
Okay.

1951
01:47:20.947 --> 01:47:25.830
On the right-hand side, you're seeing the velocity field at time t equals zero.

1952
01:47:25.850 --> 01:47:27.270
So this is what you're starting out with.

1953
01:47:27.611 --> 01:47:31.533
It's a bunch of sines and cosines for the x velocity.

1954
01:47:31.793 --> 01:47:34.154
It's a bunch of sines and cosines for the y velocity.

1955
01:47:34.874 --> 01:47:40.217
And it's zero on the z velocity, meaning all of the particles are only moving in the xy plane.

1956
01:47:40.617 --> 01:47:41.758
No one's going up and down.

1957
01:47:42.038 --> 01:47:43.939
Everything is moving side to side.

1958
01:47:43.979 --> 01:47:44.439
Flatland.

1959
01:47:44.499 --> 01:47:44.919
Flatland.

1960
01:47:45.020 --> 01:47:45.760
In flatland, right?

1961
01:47:45.980 --> 01:47:47.201
But there's bunches of flatland.

1962
01:47:47.221 --> 01:47:49.582
Like, it can still be a 3D box, but like...

1963
01:47:50.563 --> 01:47:56.707
The stuff at the top of the box is still going side to side, and the stuff at the bottom of the box is still going side to side.

1964
01:47:56.747 --> 01:47:57.648
That's how we start out with.

1965
01:47:58.028 --> 01:48:00.870
And crucially, all of these terms have sines and cosines.

1966
01:48:01.351 --> 01:48:03.793
So what that amounts to is you have vortices.

1967
01:48:04.583 --> 01:48:04.823
Okay?

1968
01:48:05.163 --> 01:48:08.684
You have a... Sines and cosines always means... circles.

1969
01:48:09.305 --> 01:48:10.565
Okay?

1970
01:48:10.785 --> 01:48:13.906
If you come out with one thing from this podcast, sines and cosines mean circles.

1971
01:48:14.266 --> 01:48:18.208
But in this case, what's happening is, you've got these vortices that are circular.

1972
01:48:18.548 --> 01:48:18.808
Okay?

1973
01:48:18.868 --> 01:48:21.509
And so all the particles are moving around in circles like this.

1974
01:48:21.549 --> 01:48:23.029
Like the ones up top are moving like this.

1975
01:48:23.470 --> 01:48:24.570
Down here, they're moving like this.

1976
01:48:24.610 --> 01:48:25.670
Down here, they're moving like this.

1977
01:48:25.830 --> 01:48:26.871
Imagine a 3D box.

1978
01:48:27.251 --> 01:48:29.892
And now I press play... on this thing.

1979
01:48:30.532 --> 01:48:31.813
What is going to happen?

1980
01:48:32.853 --> 01:48:33.093
Well...

1981
01:48:34.189 --> 01:48:34.929
Let's see what happens.

1982
01:48:36.530 --> 01:48:40.671
We've got a bunch of math here that I'm going to very quickly go through.

1983
01:48:41.691 --> 01:48:41.971
Okay?

1984
01:48:42.852 --> 01:48:48.973
For those who are not watching, my eyes just expanded to a very large size.

1985
01:48:49.153 --> 01:48:51.754
Yeah, but it's not that bad.

1986
01:48:51.874 --> 01:48:53.675
And I'll tell you what to focus on.

1987
01:48:53.755 --> 01:48:54.015
Okay.

1988
01:48:54.055 --> 01:48:54.295
Okay?

1989
01:48:55.275 --> 01:48:57.236
On the left-hand side is our initial conditions.

1990
01:48:57.396 --> 01:48:58.737
These are our sines and cosines.

1991
01:48:59.098 --> 01:49:02.420
It's just sine x and cosine y, cosine z.

1992
01:49:02.720 --> 01:49:06.582
So this tells you that the size of my vortex is 2 pi.

1993
01:49:07.749 --> 01:49:08.209
in size.

1994
01:49:08.430 --> 01:49:08.610
Okay.

1995
01:49:08.730 --> 01:49:08.990
Okay?

1996
01:49:09.691 --> 01:49:14.114
Now I apply the convective term to these equations, right?

1997
01:49:14.354 --> 01:49:16.396
These equations tell me what the velocity field is.

1998
01:49:16.816 --> 01:49:20.258
I can find the gradient of the velocity field and I can multiply it by itself.

1999
01:49:20.819 --> 01:49:28.965
And then I'll leave this as an exercise to the reader or in this case, the podcast listener, as all of the famous science and math textbooks do.

2000
01:49:29.765 --> 01:49:34.849
But you can go through and do the derivatives and then you'll have to remember your double angle identities and trig.

2001
01:49:35.330 --> 01:49:36.791
But what ends up happening is

2002
01:49:37.882 --> 01:49:43.748
those derivatives are going to create terms with a sine 2x and a cosine 2x.

2003
01:49:44.169 --> 01:49:48.373
That's going to percolate into the pressure terms.

2004
01:49:48.874 --> 01:49:50.936
You're going to get sine 2x and cosine 2x.

2005
01:49:50.956 --> 01:49:55.882
And one thing that you should remember is sine x,

2006
01:49:57.271 --> 01:49:57.771
is a wave.

2007
01:49:58.632 --> 01:50:01.213
Sine 2x is that same wave squished.

2008
01:50:01.693 --> 01:50:04.755
You're creating smaller and smaller waves.

2009
01:50:04.955 --> 01:50:06.516
The wavelength is getting smaller.

2010
01:50:06.576 --> 01:50:07.477
The oscillation.

2011
01:50:08.137 --> 01:50:08.877
Is getting squished.

2012
01:50:09.458 --> 01:50:09.698
Okay?

2013
01:50:09.978 --> 01:50:12.359
Now what does that do to our structure of the vortex?

2014
01:50:12.919 --> 01:50:13.660
What happens is...

2015
01:50:16.532 --> 01:50:18.654
Remember, before I told you that, yeah, that's the one.

2016
01:50:19.014 --> 01:50:24.359
So remember, before I told you that the z component of my velocities was zero.

2017
01:50:24.619 --> 01:50:27.802
Everything was happening in side to side, right?

2018
01:50:28.203 --> 01:50:40.474
But now, because of that convective term, and because of those higher wavelengths, sorry, I should say smaller wavelengths, there's going to be picked up z velocities.

2019
01:50:41.430 --> 01:50:43.953
There's going to be interactions that are happening in the Z direction.

2020
01:50:43.973 --> 01:50:46.576
You're going to push up stuff and you're going to push down stuff.

2021
01:50:46.977 --> 01:50:54.927
And the way you push up and down, those structures are going to be half the size of your original vortex.

2022
01:50:55.388 --> 01:50:57.290
And this is coming from the squishing.

2023
01:50:57.410 --> 01:50:57.690
Mm-hmm.

2024
01:50:58.150 --> 01:51:02.492
It's coming from the squishing and the fact that the squishing is forcing stuff into the third dimension.

2025
01:51:03.132 --> 01:51:05.613
So on the left-hand side, you see these vortices.

2026
01:51:05.633 --> 01:51:08.294
Those blobs sort of represent where the vortices are.

2027
01:51:08.334 --> 01:51:11.535
You can imagine the orange blob is a vortex going in one direction.

2028
01:51:11.575 --> 01:51:13.896
The blue blob is a vortex going in the other direction.

2029
01:51:14.196 --> 01:51:16.217
But all of those vortices are X and Y.

2030
01:51:16.777 --> 01:51:17.377
Okay?

2031
01:51:17.517 --> 01:51:19.018
They're revolving in this way.

2032
01:51:19.318 --> 01:51:21.639
And they sort of have, for lack of a better term,

2033
01:51:22.793 --> 01:51:26.034
a consistent, non-disturbed structure.

2034
01:51:26.054 --> 01:51:27.914
Yeah, and that's time t equals zero.

2035
01:51:27.934 --> 01:51:29.015
Like, that's what we started with.

2036
01:51:29.455 --> 01:51:35.896
And when we press play, all of a sudden you start getting these vortices, you're getting these interactions between the vortices up and down.

2037
01:51:36.957 --> 01:51:47.419
And the interactions, those structures that are created in that interaction, have a smaller and smaller size because of that convective term.

2038
01:51:47.579 --> 01:51:48.019
Okay?

2039
01:51:48.579 --> 01:51:49.419
I see, I see.

2040
01:51:49.479 --> 01:51:51.460
And if we continue this ad infinitum,

2041
01:51:52.703 --> 01:51:56.027
It's going to get even smaller and even smaller and even smaller.

2042
01:51:56.547 --> 01:52:01.453
To at some point, you would think, you would think, right, that like it's just going to get smaller and smaller to infinity.

2043
01:52:01.473 --> 01:52:05.017
You're going to get smaller and infinitesimally smaller structures.

2044
01:52:05.578 --> 01:52:06.359
That's not the case.

2045
01:52:06.799 --> 01:52:09.742
Viscosity at some point wins out in this scenario.

2046
01:52:10.103 --> 01:52:12.245
So coming back to how we started this.

2047
01:52:12.986 --> 01:52:17.509
We're talking about the viscosity term on the right side of our Navier-Stokes equation.

2048
01:52:18.310 --> 01:52:23.493
And then we're talking about the convective term on the left side of our Navier-Stokes equation.

2049
01:52:24.154 --> 01:52:28.257
And these two things are in competition as we looked at in this example.

2050
01:52:28.837 --> 01:52:34.121
And at time equals zero, when we have the blobs on the top left,

2051
01:52:34.841 --> 01:52:41.725
And as time goes by, the left side of the equation is winning.

2052
01:52:41.985 --> 01:52:42.566
Is winning, yeah.

2053
01:52:42.666 --> 01:52:43.126
Effectively.

2054
01:52:43.146 --> 01:52:47.748
Effectively, because it's creating smaller and smaller structure that's leaking into the z direction.

2055
01:52:48.269 --> 01:52:56.834
And so our sort of nice discrete blob structures at the top become a more chaotic system as we get these smaller and smaller structures.

2056
01:52:56.854 --> 01:52:56.994
Yes.

2057
01:52:57.674 --> 01:53:03.537
But at some point, and the fundamental question is, does this go all the way to infinity in smallness?

2058
01:53:03.557 --> 01:53:06.638
And do I get smaller and smaller structures?

2059
01:53:06.919 --> 01:53:08.960
Is the convective term the goat?

2060
01:53:09.580 --> 01:53:10.600
And does it always win?

2061
01:53:11.381 --> 01:53:12.761
And what we're saying is no.

2062
01:53:13.001 --> 01:53:13.722
In this case, no.

2063
01:53:13.922 --> 01:53:20.185
In this case, at some point, the viscosity starts to matter once you get to a sufficiently small scale.

2064
01:53:20.345 --> 01:53:29.127
Yes, because for technical audiences, the convective term goes up kind of linearly, but it grows very fast in the beginning.

2065
01:53:29.547 --> 01:53:33.207
The viscosity term grows like the square.

2066
01:53:33.747 --> 01:53:38.328
So at some point it's going to catch up and it's going to diffuse everything and then everything will just sort of be chill.

2067
01:53:38.448 --> 01:53:40.549
Yeah, that makes sense.

2068
01:53:40.749 --> 01:53:44.929
So the Taylor Greene vortex, I think it puts this conflict right in front of us.

2069
01:53:45.069 --> 01:53:48.810
You've got the nonlinear term, that convective term, that takes an organized flow

2070
01:53:49.841 --> 01:53:52.662
And it creates smaller and smaller spatial structure.

2071
01:53:53.322 --> 01:54:00.025
And finally, we've got this viscosity that becomes increasingly more powerful at those smaller and smaller spatial structures.

2072
01:54:00.245 --> 01:54:04.407
Is the convective term going like this and the viscosity is going like this?

2073
01:54:04.447 --> 01:54:05.667
It's like this versus this.

2074
01:54:05.727 --> 01:54:06.508
Okay, this versus this.

2075
01:54:06.528 --> 01:54:07.768
You know what I mean?

2076
01:54:07.808 --> 01:54:12.470
So eventually, at this meeting point of structure size…

2077
01:54:12.530 --> 01:54:15.533
Yes, is when the viscosity is like, okay, that's enough.

2078
01:54:15.733 --> 01:54:16.213
That's enough.

2079
01:54:16.293 --> 01:54:16.754
That makes sense.

2080
01:54:16.954 --> 01:54:17.554
Let's calm down.

2081
01:54:17.935 --> 01:54:25.321
And so the viscosity waits for those smaller scales to convert all of that energy into just like dissipation and heat.

2082
01:54:25.581 --> 01:54:29.465
And where it happens depends on the viscosity of the fluid.

2083
01:54:29.685 --> 01:54:32.788
If the fluid is very, very viscous, it's going to cut it immediately.

2084
01:54:33.088 --> 01:54:37.272
But if it's not very viscous, it'll take some time, but it's going to get there.

2085
01:54:38.072 --> 01:54:40.613
For the Taylor Greene Vortex, it is going to get there.

2086
01:54:40.653 --> 01:54:41.454
That's fascinating.

2087
01:54:41.474 --> 01:54:41.634
Right?

2088
01:54:41.714 --> 01:54:46.096
And now in every single simulation that we've done, viscosity always wins out.

2089
01:54:46.636 --> 01:54:46.936
Okay.

2090
01:54:47.156 --> 01:54:47.416
Okay?

2091
01:54:47.736 --> 01:54:51.078
At some point, viscosity is always going to win out.

2092
01:54:51.318 --> 01:54:52.858
V for Vendetta, one would say.

2093
01:54:52.959 --> 01:54:53.439
That's right.

2094
01:54:53.959 --> 01:54:56.860
And the Millennium Problem is asking, is that actually true?

2095
01:54:57.520 --> 01:55:00.862
For every single scenario that I can come up with...

2096
01:55:01.202 --> 01:55:02.262
I just showed you two, right?

2097
01:55:02.282 --> 01:55:05.983
We showed the co-wethe flow, and then we showed this Taylor Green vortex.

2098
01:55:06.123 --> 01:55:08.044
In both of those cases, pretty fine.

2099
01:55:08.144 --> 01:55:08.884
Everything's smooth.

2100
01:55:09.104 --> 01:55:10.024
Everything's normal.

2101
01:55:10.724 --> 01:55:11.845
Things aren't going crazy.

2102
01:55:12.785 --> 01:55:15.126
So is this the idea that smoothness always wins?

2103
01:55:15.266 --> 01:55:15.546
Yes.

2104
01:55:16.086 --> 01:55:18.126
Like to describe it in another way.

2105
01:55:18.186 --> 01:55:21.127
The solution is always going to be smooth because of the viscosity.

2106
01:55:21.187 --> 01:55:22.227
Because of the viscosity.

2107
01:55:22.267 --> 01:55:23.608
Winning in this race.

2108
01:55:24.088 --> 01:55:24.188
Yeah.

2109
01:55:24.208 --> 01:55:24.328
Okay.

2110
01:55:24.468 --> 01:55:24.788
Okay.

2111
01:55:25.228 --> 01:55:26.188
Is it always going to win?

2112
01:55:26.468 --> 01:55:27.169
Is it always going to win?

2113
01:55:27.229 --> 01:55:27.449
Yes.

2114
01:55:27.569 --> 01:55:29.189
We've shown that it can win sometimes.

2115
01:55:29.349 --> 01:55:29.569
Yes.

2116
01:55:29.910 --> 01:55:31.130
And for those two times, it's going to win.

2117
01:55:31.170 --> 01:55:31.330
Sure.

2118
01:55:31.410 --> 01:55:35.171
And actually, for every single simulation that we've done, we've shown that it's going to win.

2119
01:55:36.011 --> 01:55:39.693
Every single simulation in the history of aerodynamics, we've shown that it's going to win.

2120
01:55:39.973 --> 01:55:40.213
Now...

2121
01:55:40.993 --> 01:55:43.435
that could be an artifact of the fact that it's a simulation, right?

2122
01:55:43.495 --> 01:55:45.556
And there's like discretized grids.

2123
01:55:46.037 --> 01:55:48.078
And so we can't go infinitesimally small.

2124
01:55:48.198 --> 01:55:55.563
At some point, maybe it's an artifact of the fact that like, you literally can't physically, like there's not enough compute to go down to what?

2125
01:55:55.583 --> 01:55:57.144
The Planck length or something like that, right?

2126
01:55:57.224 --> 01:55:58.665
Where it could start falling apart.

2127
01:55:58.705 --> 01:56:00.526
Yeah, like, so perhaps that's that.

2128
01:56:00.567 --> 01:56:02.268
And that's what the Millennium Problem is asking.

2129
01:56:02.628 --> 01:56:03.729
It's asking something stronger.

2130
01:56:03.769 --> 01:56:09.873
It's saying for every single initial condition that I could come up with,

2131
01:56:11.292 --> 01:56:13.133
Is viscosity always going to win?

2132
01:56:13.734 --> 01:56:16.635
This is the is smoothness global idea.

2133
01:56:17.356 --> 01:56:34.927
And that actually, it's so funny because now it's, it kind of reminds me a little bit of, was it the Riemann hypothesis episode where we talked about the non- Trivial zeros.

2134
01:56:34.947 --> 01:56:36.108
The non-trivial zeros.

2135
01:56:36.168 --> 01:56:37.749
And it's similarly like, you know,

2136
01:56:38.169 --> 01:56:43.754
We want to take the question to its ultimate... Yeah, is every single zero on the critical line?

2137
01:56:43.774 --> 01:56:44.555
On the critical line.

2138
01:56:44.795 --> 01:56:45.455
Every single one.

2139
01:56:45.475 --> 01:56:46.476
No, every single one.

2140
01:56:46.536 --> 01:56:48.038
Because you can still have 99.99.

2141
01:56:48.318 --> 01:56:50.520
You can even have 100, as we were saying.

2142
01:56:50.860 --> 01:56:55.304
And that doesn't mean... You can have 100% and still have one off.

2143
01:56:55.644 --> 01:57:02.390
And anywhere in all possibility that one is off, it means... And that's what mathematicians are worried about, right?

2144
01:57:02.470 --> 01:57:03.691
It's like, no, no, no, no.

2145
01:57:03.831 --> 01:57:05.353
In every universe...

2146
01:57:05.993 --> 01:57:06.253
Right?

2147
01:57:06.353 --> 01:57:08.575
That Doctor Strange has ever visited.

2148
01:57:10.116 --> 01:57:10.757
Is this true?

2149
01:57:11.798 --> 01:57:13.839
It feels similarly in that direction.

2150
01:57:13.880 --> 01:57:16.782
Yeah, it's a very mathematical thing to worry about.

2151
01:57:16.802 --> 01:57:18.563
Now I understand the mindset.

2152
01:57:18.723 --> 01:57:18.944
Yeah.

2153
01:57:19.644 --> 01:57:23.567
Now, crucially, this is now an open problem only in 3D.

2154
01:57:25.263 --> 01:57:27.865
In 2D, there was a very famous mathematician.

2155
01:57:27.905 --> 01:57:29.386
She was a Soviet mathematician.

2156
01:57:30.347 --> 01:57:33.108
Olga Ladijenskaya.

2157
01:57:33.709 --> 01:57:34.890
Olga Ladijenskaya.

2158
01:57:35.210 --> 01:57:44.516
She wrote this 1969 book, The Mathematical Theory of Viscous Incompressible Flow, where she showed that in two dimensions, yes, yes,

2159
01:57:45.244 --> 01:57:49.066
every single initial condition guaranteed viscosity is going to win.

2160
01:57:49.306 --> 01:57:50.566
If we're in flatland.

2161
01:57:51.107 --> 01:57:51.307
Yes.

2162
01:57:51.467 --> 01:57:54.208
If we're in flatland and we've got like fluids flowing in flatland.

2163
01:57:54.248 --> 01:57:56.709
And there's no Z up, Z down that we just described.

2164
01:57:56.869 --> 01:57:59.210
As we saw actually, like even in the Taylor green vortex.

2165
01:57:59.250 --> 01:57:59.650
Oh, okay.

2166
01:57:59.750 --> 01:57:59.911
Right.

2167
01:57:59.931 --> 01:58:10.135
We saw that like the Z dimension was the one that was causing that weirdness because the flow was escaping into the Z dimension and then creating those smaller and smaller structures.

2168
01:58:10.275 --> 01:58:14.457
So Olga showed in 1969 in her very famous book that,

2169
01:58:14.833 --> 01:58:15.213
Indeed.

2170
01:58:15.594 --> 01:58:17.255
In two dimensions, it's totally fine.

2171
01:58:17.835 --> 01:58:25.400
Every single starting point that you could ever think of, you're going to come up with a smooth solution.

2172
01:58:26.901 --> 01:58:27.161
Okay.

2173
01:58:27.261 --> 01:58:30.944
So for 2D, the Navier-Stokes global smoothness is known.

2174
01:58:31.244 --> 01:58:31.724
It's fine.

2175
01:58:31.924 --> 01:58:33.285
That's not the question we're asking.

2176
01:58:33.465 --> 01:58:33.966
Yeah.

2177
01:58:34.026 --> 01:58:34.526
For 3D.

2178
01:58:34.706 --> 01:58:35.867
For 3D is...

2179
01:58:36.247 --> 01:58:36.927
It is not known.

2180
01:58:37.748 --> 01:58:39.489
And is this... And I just have a quick question.

2181
01:58:39.829 --> 01:58:43.632
So when I see the term blow up... Mm-hmm.

2182
01:58:44.288 --> 01:58:46.230
as it relates to how people describe this.

2183
01:58:46.310 --> 01:58:48.852
This is what we're... Is this kind of what we're talking about?

2184
01:58:48.872 --> 01:58:51.694
It's like, does the smoothness...

2185
01:58:53.475 --> 01:58:54.676
Does viscosity lose?

2186
01:58:55.077 --> 01:58:55.797
Yes.

2187
01:58:55.997 --> 01:58:57.178
Which is effectively... Blow up.

2188
01:58:57.298 --> 01:58:57.799
Yeah.

2189
01:58:57.819 --> 01:58:58.940
Does viscosity lose?

2190
01:58:59.000 --> 01:59:00.861
Because if viscosity loses, then that can...

2191
01:59:00.961 --> 01:59:09.288
Effective term is going to keep going on this runaway reaction to create smaller and smaller structures of higher and higher velocities...

2192
01:59:09.568 --> 01:59:12.989
And at some point you're going to get like stuff that's moving at infinite speed.

2193
01:59:13.990 --> 01:59:14.250
Right?

2194
01:59:15.250 --> 01:59:18.631
It's kind of like, is there a singularity in fluids?

2195
01:59:18.771 --> 01:59:19.191
Not really.

2196
01:59:19.411 --> 01:59:19.752
But like.

2197
01:59:19.852 --> 01:59:20.732
Well, no, but no, no, no.

2198
01:59:20.752 --> 01:59:22.352
That's literally what they're asking.

2199
01:59:22.813 --> 01:59:23.853
Is there a singularity?

2200
01:59:24.193 --> 01:59:26.134
Like, can I create a singularity?

2201
01:59:26.214 --> 01:59:27.774
That's what they're asking.

2202
01:59:27.814 --> 01:59:28.374
Okay, fair enough.

2203
01:59:28.475 --> 01:59:29.755
That's exactly what they're asking.

2204
01:59:30.035 --> 01:59:32.155
Now, why is it so hard in three dimensions?

2205
01:59:32.175 --> 01:59:36.156
Well, we already saw, like, kind of a clue to that with the Taylor Greene vortex, right?

2206
01:59:36.476 --> 01:59:38.917
It's the fact that, like, stuff can move up and down.

2207
01:59:39.257 --> 01:59:42.437
And crucially, that had to do with a vortex that I had started, right?

2208
01:59:42.457 --> 01:59:43.957
The initial condition was a vortex.

2209
01:59:44.237 --> 01:59:53.059
The sines and cosines described stuff going around in a circle, and we had these mini tornadoes inside of our thing that were interacting with one another and leaking into one another.

2210
01:59:54.439 --> 01:59:57.060
Vorticity is the problem in 3D, okay?

2211
01:59:57.100 --> 01:59:58.360
Vorticity is the curl, right?

2212
01:59:59.420 --> 02:00:09.267
For those who have taken calculus, the curl of the vector field, that del operator is back again, but this time we're taking a cross product of the del operator with the velocity field.

2213
02:00:09.747 --> 02:00:13.109
It's just a way of saying, is the velocity field curling?

2214
02:00:13.409 --> 02:00:16.191
Like if I put a pinwheel in the velocity field, is it going to turn?

2215
02:00:16.752 --> 02:00:19.573
If it's going to turn, then that means that there's a curl that's happening over there.

2216
02:00:19.954 --> 02:00:20.194
Okay.

2217
02:00:20.474 --> 02:00:24.797
Now in two dimensions, vorticity has nowhere to go.

2218
02:00:26.073 --> 02:00:26.293
Right.

2219
02:00:26.654 --> 02:00:26.874
Right?

2220
02:00:26.934 --> 02:00:30.117
It's just spinning, and then it's got to, like, dissipate in the flatland.

2221
02:00:30.197 --> 02:00:30.438
Yeah.

2222
02:00:31.238 --> 02:00:34.622
In three dimensions, you can have something called vortex stretching.

2223
02:00:35.963 --> 02:00:36.223
Okay?

2224
02:00:36.484 --> 02:00:37.645
No, this is in two dimensions.

2225
02:00:38.005 --> 02:00:42.490
You've got vorticity that's, like, kind of just spreading out and, like, going out about.

2226
02:00:42.850 --> 02:00:43.251
Like...

2227
02:00:44.303 --> 02:00:47.885
You know, it's a small vortex, then it becomes a bigger and bigger vortex, right?

2228
02:00:48.145 --> 02:00:49.906
And that's all it can do in two dimensions.

2229
02:00:49.926 --> 02:00:50.726
Because it's nowhere to go.

2230
02:00:50.806 --> 02:00:51.507
It's got nowhere to go.

2231
02:00:51.587 --> 02:00:54.948
In three dimensions, you get something called vortex stretching, okay?

2232
02:00:55.408 --> 02:00:56.569
And this could be a problem.

2233
02:00:57.730 --> 02:01:02.292
Imagine a vortex that has a certain size, it's got a certain radius, like a hurricane.

2234
02:01:02.752 --> 02:01:04.873
That's our... Yeah, let's say a hurricane.

2235
02:01:05.293 --> 02:01:12.237
It's quite big in space, and it's moving, but the velocity is not that high, okay?

2236
02:01:12.257 --> 02:01:12.457
Okay.

2237
02:01:13.402 --> 02:01:17.266
And now that hurricane, all of that energy gets squished into a tornado.

2238
02:01:17.646 --> 02:01:18.307
Hmm.

2239
02:01:19.203 --> 02:01:19.683
What's going to happen?

2240
02:01:19.703 --> 02:01:27.346
Well, the tornado is going to get very, very tall because all of that energy from the hurricane has to... Divergence has to equal zero.

2241
02:01:27.386 --> 02:01:27.586
Yeah.

2242
02:01:28.086 --> 02:01:28.706
Very good.

2243
02:01:28.966 --> 02:01:29.787
This is exactly it.

2244
02:01:29.807 --> 02:01:31.227
It has to equal zero.

2245
02:01:31.287 --> 02:01:32.028
So it's got to go somewhere.

2246
02:01:32.048 --> 02:01:32.888
It's got to go somewhere.

2247
02:01:32.948 --> 02:01:34.648
So it's going to go up, right?

2248
02:01:34.869 --> 02:01:38.390
But because it's getting tall, again, the radius is going to decrease.

2249
02:01:38.710 --> 02:01:44.412
But if the radius decreases, now we have another principle of physics, which is the conservation of angular momentum.

2250
02:01:44.872 --> 02:01:46.673
And so just like, you know...

2251
02:01:47.759 --> 02:01:53.902
In the figure skaters, when they're twirling around, they pull their arms in, they spin faster.

2252
02:01:54.182 --> 02:01:55.203
Same thing's going to happen here.

2253
02:01:55.863 --> 02:01:57.204
The thing is going to spin faster.

2254
02:01:57.844 --> 02:02:05.168
And this kind of reminds me in a different context of our sideways wine bottle and then it getting narrower.

2255
02:02:05.288 --> 02:02:06.788
And so it increased the acceleration.

2256
02:02:07.109 --> 02:02:09.150
Just conceptually, there's a sort of similar thing.

2257
02:02:09.190 --> 02:02:10.470
Yeah, you're packing stuff in here.

2258
02:02:10.510 --> 02:02:14.052
But here it's the vorticity that's increasing because of the conservation of angular momentum.

2259
02:02:14.432 --> 02:02:17.413
Again, it's kind of similar because there's something being conserved.

2260
02:02:17.773 --> 02:02:19.273
In that case, it was the amount of fluid.

2261
02:02:19.553 --> 02:02:21.354
In this case, it's the amount of angular momentum.

2262
02:02:21.374 --> 02:02:21.854
Makes sense.

2263
02:02:21.874 --> 02:02:26.375
And this is called vortex stretching because you're taking a vortex, you're stretching it up.

2264
02:02:26.675 --> 02:02:28.195
But as you stretch it up, you thin it out.

2265
02:02:28.515 --> 02:02:32.276
And if you thin it out, it's like the figure skater bringing in her arms.

2266
02:02:32.776 --> 02:02:34.437
The thing is going to move faster and faster.

2267
02:02:34.837 --> 02:02:40.678
And this kind of relates to what we just talked about in the Taylor Green vortex, why we're now leaking into the...

2268
02:02:41.138 --> 02:02:42.439
Into those higher dimensions, right?

2269
02:02:42.459 --> 02:02:42.679
Right.

2270
02:02:42.719 --> 02:02:49.561
And the question is, maybe with this vortex stretching, you can actually create a kind of singularity.

2271
02:02:50.542 --> 02:02:50.802
Okay?

2272
02:02:51.182 --> 02:02:53.903
Now, let's get into what we mean by singularity in the first place.

2273
02:02:54.163 --> 02:02:56.744
We had already done our first toy differential equation, right?

2274
02:02:56.764 --> 02:02:56.964
Yes.

2275
02:02:57.124 --> 02:02:59.185
Which is the compound interest differential equation.

2276
02:02:59.645 --> 02:03:01.166
Now, this thing does go to infinity.

2277
02:03:01.186 --> 02:03:01.426
Okay.

2278
02:03:02.012 --> 02:03:02.212
right?

2279
02:03:02.833 --> 02:03:09.760
I mean, it's E to the X or E to the T. For infinite time, I'll get an infinite amount of money.

2280
02:03:10.621 --> 02:03:17.688
If I wait long enough, even if I have a dollar in my bank account, that dollar is going to become infinity if I wait long enough, right?

2281
02:03:18.889 --> 02:03:20.971
It just turns out maybe AI is going to end the

2282
02:03:23.553 --> 02:03:24.795
That's not a real singularity.

2283
02:03:24.875 --> 02:03:27.898
A real singularity is an infinite time.

2284
02:03:28.038 --> 02:03:30.901
I shouldn't say infinite because that sounds like infinite.

2285
02:03:31.061 --> 02:03:37.847
So I should say before in finite time, can I get to infinity?

2286
02:03:37.867 --> 02:03:38.047
Yeah.

2287
02:03:44.172 --> 02:03:46.213
In some constrained amount of time.

2288
02:03:46.233 --> 02:03:46.633
There we go.

2289
02:03:46.714 --> 02:03:46.954
Yes.

2290
02:03:47.614 --> 02:03:49.715
For our listeners out there, that'll be good.

2291
02:03:49.835 --> 02:03:51.916
In a constrained amount of time, can I get to infinity?

2292
02:03:52.237 --> 02:03:54.218
Here's another differential equation that does that.

2293
02:03:54.918 --> 02:03:59.201
Here, the derivative, my derivative with respect to time, is equal to y squared.

2294
02:03:59.441 --> 02:04:01.182
So dy dt is equal to y squared.

2295
02:04:02.182 --> 02:04:05.884
Initial condition, you start at 1 when time is 0.

2296
02:04:07.205 --> 02:04:12.148
And the solution to that is 1 over 1 minus t. If I plug in t equals 1...

2297
02:04:13.238 --> 02:04:13.979
I get infinity.

2298
02:04:14.879 --> 02:04:15.060
Right?

2299
02:04:15.580 --> 02:04:21.425
So this is a differential equation that creates a singularity, but the time is not infinity.

2300
02:04:21.465 --> 02:04:22.286
The time is just one.

2301
02:04:22.446 --> 02:04:24.808
At one second, I have a singularity.

2302
02:04:24.848 --> 02:04:25.709
You reach the singularity.

2303
02:04:26.049 --> 02:04:27.250
Another thing I want to point out here.

2304
02:04:28.471 --> 02:04:30.273
What's special about this differential equation?

2305
02:04:30.773 --> 02:04:31.614
It is nonlinear.

2306
02:04:32.515 --> 02:04:32.915
You see?

2307
02:04:33.676 --> 02:04:34.556
It's y squared.

2308
02:04:34.616 --> 02:04:35.157
Yeah.

2309
02:04:35.787 --> 02:04:38.511
It's the simplest nonlinear differential equation.

2310
02:04:38.611 --> 02:04:41.635
And already I am getting a singularity, right?

2311
02:04:41.936 --> 02:04:44.079
The other one was just dy dt equals y.

2312
02:04:44.239 --> 02:04:45.280
That's a linear equation.

2313
02:04:45.601 --> 02:04:46.743
I can just add up the terms.

2314
02:04:47.956 --> 02:04:56.679
But here, the non-linearity is creating that cascade that is creating a singularity before time goes to infinity, at a finite time.

2315
02:04:57.259 --> 02:05:00.020
So it's conceivable to extrapolate from this.

2316
02:05:00.220 --> 02:05:00.460
Yeah.

2317
02:05:00.700 --> 02:05:09.902
It would be conceivable based on how we've constructed the non-linearity of trying to look at this incompressible fluid and look at a vector field.

2318
02:05:10.002 --> 02:05:11.083
Yeah, it's like there's a chance.

2319
02:05:11.363 --> 02:05:14.484
There's a chance because we got a lot of non-linearity going on.

2320
02:05:14.504 --> 02:05:15.004
Yeah, yeah.

2321
02:05:15.224 --> 02:05:15.684
It's just that...

2322
02:05:16.550 --> 02:05:18.611
There's the viscosity term, right?

2323
02:05:18.731 --> 02:05:21.633
In this one that I showed you, there's no smoothing.

2324
02:05:21.653 --> 02:05:22.894
There's no counteracting.

2325
02:05:22.914 --> 02:05:23.495
It's just going.

2326
02:05:23.595 --> 02:05:23.975
Right?

2327
02:05:24.075 --> 02:05:27.057
But in the Navier-Stokes, and that's why it's so interesting.

2328
02:05:27.097 --> 02:05:32.221
Because the Navier-Stokes has this non-linearity, but it also has this smoothing term.

2329
02:05:32.301 --> 02:05:34.122
It has this counterforce, which is the viscosity.

2330
02:05:34.162 --> 02:05:36.043
And so you're asking, who's going to win?

2331
02:05:36.423 --> 02:05:38.185
And are they always going to win?

2332
02:05:38.285 --> 02:05:40.866
So far, it seems viscosity is always won.

2333
02:05:41.006 --> 02:05:43.208
That makes so much sense now.

2334
02:05:43.408 --> 02:05:44.749
That's so good.

2335
02:05:44.949 --> 02:05:45.229
Right?

2336
02:05:45.309 --> 02:05:46.050
That's so good.

2337
02:05:46.250 --> 02:05:46.850
It's quite nice.

2338
02:05:46.870 --> 02:05:47.830
Yeah, it's quite nice.

2339
02:05:48.171 --> 02:05:50.331
So it is a very interesting problem.

2340
02:05:50.351 --> 02:05:56.433
Yeah, no, because you can understand – there's good arguments for both sides.

2341
02:05:57.173 --> 02:06:01.154
One in which viscosity always wins feels true.

2342
02:06:01.334 --> 02:06:02.594
Yeah, seems legit.

2343
02:06:02.694 --> 02:06:04.815
Because we've been able to show that.

2344
02:06:06.597 --> 02:06:07.657
Over and over again.

2345
02:06:07.997 --> 02:06:10.358
In a variety of contexts as being true.

2346
02:06:10.578 --> 02:06:11.158
Yeah.

2347
02:06:11.218 --> 02:06:14.559
But being fundamentally true... Is another thing.

2348
02:06:14.679 --> 02:06:15.620
Is another thing.

2349
02:06:15.900 --> 02:06:29.864
And because of, again, the nonlinear nature of the convective term in this case, that is challenging the viscosity as the counterforce, there's an argument that both of them have a reason that they could win.

2350
02:06:30.404 --> 02:06:31.704
And in finite time.

2351
02:06:31.724 --> 02:06:32.464
In finite time.

2352
02:06:32.785 --> 02:06:33.225
Exactly.

2353
02:06:33.345 --> 02:06:33.805
And so...

2354
02:06:34.705 --> 02:06:36.106
It's an interesting problem.

2355
02:06:36.126 --> 02:06:36.366
I like that.

2356
02:06:36.386 --> 02:06:37.987
I have to give it to him, right?

2357
02:06:38.127 --> 02:06:41.448
It's something that definitely makes you think.

2358
02:06:41.788 --> 02:06:45.310
And there are other things in physics that have singularities in finite time.

2359
02:06:45.430 --> 02:06:50.292
I mean, the most famous one would probably be a black hole as a result of the Einstein field equations.

2360
02:06:50.593 --> 02:07:01.638
The Einstein field equations are also very famously a nonlinear differential equation because you've got these tensors, and these tensors are multiplied by,

2361
02:07:02.475 --> 02:07:05.899
to themselves and also multiplied to the reciprocal of themselves twice over.

2362
02:07:06.379 --> 02:07:08.862
And so you get these squared terms.

2363
02:07:08.902 --> 02:07:16.750
For example, in a very simple case, the curvature of space and time causes gravity, right?

2364
02:07:17.311 --> 02:07:21.696
But gravity itself has an energy that causes the curvature of space and time.

2365
02:07:23.453 --> 02:07:23.934
Right?

2366
02:07:24.174 --> 02:07:24.715
So what?

2367
02:07:25.015 --> 02:07:31.765
And so now you get these nonlinear actions on space and time that create the richness of the phenomenon that we see.

2368
02:07:32.105 --> 02:07:36.031
And on the right-hand side, I mentioned Lake Geneva.

2369
02:07:36.892 --> 02:07:38.915
I was actually at the Yerkes Observatory.

2370
02:07:39.295 --> 02:07:42.256
on Lake Geneva, which is known as the birthplace of modern astrophysics.

2371
02:07:42.556 --> 02:07:43.816
We'll do an episode on them.

2372
02:07:44.317 --> 02:07:50.558
They had a nice sculpture of Einstein's general relativity because Einstein had visited the Yerkes Observatory many times.

2373
02:07:50.918 --> 02:07:56.220
And there on the slab of marble was written out Einstein's field equation.

2374
02:07:56.240 --> 02:08:02.502
So I just wanted to show that when I received this news, I was looking at a non-linearity.

2375
02:08:03.502 --> 02:08:19.695
live and i was like damn right here we go again here we go again you know like the meme we started the episode with exactly so singularities can exist in physics right the center of a black hole is a singularity but i do have to say that in this case in the navier stokes case

2376
02:08:20.879 --> 02:08:22.901
This singularity is not a physical singularity.

2377
02:08:23.361 --> 02:08:26.684
And physicists do not care about this singularity.

2378
02:08:27.104 --> 02:08:28.305
It's purely for the math.

2379
02:08:28.465 --> 02:08:31.008
It's purely a mathematical question, okay?

2380
02:08:31.408 --> 02:08:35.411
Physicists do not care about the Millennium Problem in Navier-Stokes.

2381
02:08:36.032 --> 02:08:41.176
Because fluids are never going to achieve an actual physical singularity.

2382
02:08:41.777 --> 02:08:44.919
There are plenty of cases where the Navier-Stokes equations don't even apply.

2383
02:08:44.939 --> 02:08:46.921
And I've got some examples over here.

2384
02:08:47.401 --> 02:08:48.302
I mean, for example...

2385
02:08:49.813 --> 02:08:51.775
hypersonics, and supersonics even.

2386
02:08:52.135 --> 02:08:53.737
Navier-Stokes equations don't apply.

2387
02:08:54.057 --> 02:08:56.560
Navier-Stokes equations are both of those equations.

2388
02:08:57.300 --> 02:09:03.066
One of the key ones is that second one, the incompressibility, the fact that the density is always constant.

2389
02:09:03.427 --> 02:09:08.412
Well, when you're approaching the sound barrier, all of a sudden air becomes compressible.

2390
02:09:08.592 --> 02:09:08.812
Right.

2391
02:09:09.172 --> 02:09:13.437
The sound wave is catching up to the sound wave in front of you.

2392
02:09:14.017 --> 02:09:16.278
And that is going to compress the air.

2393
02:09:16.738 --> 02:09:24.219
So that's the bottom there where you get a sonic boom and you get like a cloud because like literal condensation forms around your aircraft.

2394
02:09:24.899 --> 02:09:32.801
Hypersonics is when you're going so fast that the chemical nature of the air starts mattering and you get thermal effects.

2395
02:09:33.301 --> 02:09:35.862
That's not in the Navier-Stokes equations.

2396
02:09:36.002 --> 02:09:38.302
In the middle, that's an aquaporin protein.

2397
02:09:39.002 --> 02:09:41.843
Those things are so small that they let in only water molecules.

2398
02:09:43.229 --> 02:09:48.471
At a molecular scale, velocity fields don't matter at a molecular scale.

2399
02:09:48.491 --> 02:09:49.992
Like, what are we talking about?

2400
02:09:50.012 --> 02:09:50.692
Right, right, right.

2401
02:09:50.752 --> 02:09:53.993
So again, at the molecular scale, Navier-Stokes equations don't apply.

2402
02:09:54.673 --> 02:10:02.616
And for a lot of weather patterns even, Navier-Stokes equations, at least the ones that we have up top here, are not the ones that are used.

2403
02:10:03.036 --> 02:10:10.699
We use a compressible version of Navier-Stokes equations where the air can expand and contract and change densities.

2404
02:10:11.360 --> 02:10:12.480
That's how we predict weather.

2405
02:10:13.750 --> 02:10:20.292
The millennium problem, which is the mathematical problem, has to do with these two specific equations.

2406
02:10:20.613 --> 02:10:23.254
Which are specifically for incompressible fluids.

2407
02:10:23.274 --> 02:10:23.534
Yes.

2408
02:10:23.934 --> 02:10:27.435
And so I would even say that the millennium problem has nothing to do with fluids.

2409
02:10:28.195 --> 02:10:32.777
It has to do with a vector field that is governed by this equation.

2410
02:10:33.417 --> 02:10:42.281
It has to do with an object, a mathematical object that is a vector field, that in normal circumstances would describe normal fluids.

2411
02:10:43.247 --> 02:10:49.932
But if you're trying to hunt for singularities, you have left the real world and you have entered the world of mathematics.

2412
02:10:50.252 --> 02:10:51.313
And it's totally legitimate.

2413
02:10:51.453 --> 02:10:51.993
100%.

2414
02:10:52.073 --> 02:10:52.293
Right?

2415
02:10:53.074 --> 02:10:55.576
But it's something that is outside the realm of physics.

2416
02:10:56.076 --> 02:10:58.878
And I want to make that very clear because there's a lot of hullabaloo.

2417
02:10:59.078 --> 02:11:03.121
And I think there's a lot of stories out there being like, oh, this is like fluids, right?

2418
02:11:03.401 --> 02:11:05.242
Can a fluid achieve singularity?

2419
02:11:05.322 --> 02:11:05.503
No.

2420
02:11:05.523 --> 02:11:05.583
No.

2421
02:11:06.283 --> 02:11:07.824
Actually, the answer is no, right?

2422
02:11:07.984 --> 02:11:08.644
Infinite speed.

2423
02:11:08.704 --> 02:11:10.565
Again, Einstein would be rolling over his grave.

2424
02:11:10.585 --> 02:11:19.609
You can't have infinite speed because as like if you have a fluid, right, that is like going faster and faster and faster, then all of a sudden Navier-Stokes better account for relativity.

2425
02:11:20.030 --> 02:11:31.115
And the fact that the mass of the fluid, like that density term, is going to have a Lorentz factor in there or something because the mass of the fluid is going to infinity the closer I approach the speed of light.

2426
02:11:31.795 --> 02:11:33.016
I can't go to infinity.

2427
02:11:33.597 --> 02:11:42.864
So all I'm saying is this is a mathematics question, completely legitimate, but let's not confuse this for something that is physically relevant.

2428
02:11:43.725 --> 02:11:44.226
It is not.

2429
02:11:44.626 --> 02:11:51.332
And I think we've now built up a full understanding to be able to specify...

2430
02:11:52.272 --> 02:12:09.744
What exactly the Millennium Prize problem as it relates to Navier Stokes is specifically asking about incompressible fluids and sort of do you get blow up in finite time of an incompressible fluid?

2431
02:12:10.304 --> 02:12:25.798
And in this context, what we're saying is we're hunting for singularities in these mathematical constructs that don't apply to the real world, but still have a fundamental value to our mathematical understanding of the world of just math.

2432
02:12:26.278 --> 02:12:31.983
Which, again, then can bleed into, like I was mentioning earlier, two ways to think about other problems.

2433
02:12:32.344 --> 02:12:37.028
Because it's this human understanding that creates – look at how we went from Newton's –

2434
02:12:38.145 --> 02:12:46.208
Second law of motion to the Navier-Stokes equations because it created a framework and we said, can we apply this framework to fluids?

2435
02:12:46.308 --> 02:12:46.528
Yes.

2436
02:12:46.728 --> 02:12:49.309
And then it created this whole centuries-long journey.

2437
02:12:49.329 --> 02:12:59.772
Yeah, I mean, a lot of the stuff that led up to the Navier-Stokes stuff had to do with just analyzing what vector fields are like and what it means to do calculus on vector fields.

2438
02:13:00.072 --> 02:13:01.413
Like things like Stokes' theorem.

2439
02:13:02.053 --> 02:13:04.496
That's from the Navier-Stokes guy, right?

2440
02:13:04.516 --> 02:13:09.623
That's a mathematics theorem, Stokes theorem, that like then, oh, it like works here, right?

2441
02:13:10.043 --> 02:13:13.748
So definitely something that is super interesting.

2442
02:13:14.008 --> 02:13:17.112
And Terry Tao actually said that like the Navier-Stokes – Oh, we're calling him Terry now.

2443
02:13:17.132 --> 02:13:17.673
Look at this guy.

2444
02:13:17.793 --> 02:13:19.015
Oh, my buddy Terry.

2445
02:13:20.123 --> 02:13:20.944
We need to edit that out.

2446
02:13:21.004 --> 02:13:21.524
Hey, yo.

2447
02:13:21.704 --> 02:13:25.587
If somebody... All right, Terence Tao.

2448
02:13:25.707 --> 02:13:26.688
No, no, no.

2449
02:13:26.808 --> 02:13:27.169
Keep it in.

2450
02:13:27.369 --> 02:13:35.015
But like Terence Tao, he said the Navier-Stokes equations are one of the simplest, super critical differential equations.

2451
02:13:35.515 --> 02:13:43.501
So in the theory of differential equations, it's very interesting to think about because you've got this nonlinear term and you've got this competition with the diffusion term, right?

2452
02:13:43.802 --> 02:13:48.205
You've got a Laplacian and you've got this like convective thing and they're competing, right?

2453
02:13:48.445 --> 02:13:51.549
And mathematically, it's a very interesting question to answer.

2454
02:13:52.430 --> 02:14:01.982
Now, there's been a lot of work since the days of Navier Stokes to try and understand the mathematical properties of this.

2455
02:14:02.254 --> 02:14:04.677
The first guy I want to talk about is Gene Lire.

2456
02:14:05.718 --> 02:14:13.687
He actually proved that fluids always possess something called weak solutions, where the total kinetic energy of the system remains bounded over time.

2457
02:14:13.967 --> 02:14:15.769
How do you compute kinetic energy?

2458
02:14:16.070 --> 02:14:19.774
Like, kinetic energy is really just 1.5 mv squared, right?

2459
02:14:19.934 --> 02:14:22.276
For, like, a normal point-like object, like...

2460
02:14:22.837 --> 02:14:25.860
You know how Newton's laws can be applied to a point-lock object, f equals ma?

2461
02:14:25.880 --> 02:14:27.501
Well, one-half mv squared.

2462
02:14:27.541 --> 02:14:29.483
For a velocity field, it's a little bit different, right?

2463
02:14:29.723 --> 02:14:45.077
What we have to do is integrate over a volume, because we've got a bunch of velocities in a volume, and we say the square of all of the velocities in that volume times one-half, that's going to tell you sort of the energy density in that volume, like the kinetic energy density in that volume.

2464
02:14:45.097 --> 02:14:46.779
Which would be, in this case, the velocity field.

2465
02:14:46.799 --> 02:14:46.919
Right.

2466
02:14:46.959 --> 02:14:48.080
Yeah, yeah, exactly.

2467
02:14:48.300 --> 02:14:54.165
And so the kinetic energy density can remain finite for these weak solutions.

2468
02:14:54.205 --> 02:14:59.850
And by weak solutions, these are stuff that don't have this strong smoothness criteria.

2469
02:15:00.270 --> 02:15:04.093
And I'm going to be honest, at this point, I'm getting into stuff that...

2470
02:15:04.153 --> 02:15:05.014
This is the math stuff.

2471
02:15:05.414 --> 02:15:10.498
This is maybe why I went on a tirade about five minutes earlier about how this is all math.

2472
02:15:10.578 --> 02:15:14.561
Because maybe there's stuff now that I'm like, okay, I don't really know, right?

2473
02:15:14.842 --> 02:15:15.022
But...

2474
02:15:16.348 --> 02:15:25.636
From what I've gathered, what he's done is loosen the idea of smoothness and show that, like, you know, imagine you've got, like, a violent storm system.

2475
02:15:26.217 --> 02:15:29.900
He showed that the total energy of that storm can't suddenly go to infinity.

2476
02:15:30.440 --> 02:15:30.680
Okay?

2477
02:15:30.981 --> 02:15:38.327
But what he can't rule out is that even though the entire energy density of that storm doesn't go to infinity, you can have small pockets of...

2478
02:15:39.083 --> 02:15:40.965
where the velocity is still going to infinity.

2479
02:15:41.305 --> 02:15:47.071
Okay, you can concentrate into an infinitely fast, like, tornado somewhere inside, and you can't rule that out.

2480
02:15:47.792 --> 02:15:51.716
So the overall energy is finite, but some local small velocity can be infinite.

2481
02:15:52.277 --> 02:15:58.423
So already it's starting to show that maybe there could be ways that the convective term wins.

2482
02:15:58.724 --> 02:15:59.004
Mm-hmm.

2483
02:15:59.765 --> 02:16:12.339
1982, Kaffarelli, Cohn, and Nuremberg, they show that if the Navier-Stokes equations for a velocity field work, then you have partial regularity.

2484
02:16:12.419 --> 02:16:16.463
What that means is, if you have singularities, if they exist at all,

2485
02:16:17.338 --> 02:16:18.798
They can't be like everywhere.

2486
02:16:19.399 --> 02:16:19.759
Okay.

2487
02:16:19.919 --> 02:16:23.460
You can't have giant chunks of fluid moving at infinite speed.

2488
02:16:23.880 --> 02:16:30.661
But what you could do is have infinitesimally small point-like singularities, kind of like at the center of a black hole.

2489
02:16:31.582 --> 02:16:32.702
And that works.

2490
02:16:33.722 --> 02:16:34.002
Okay?

2491
02:16:34.862 --> 02:16:39.884
And so the system has some fundamental limitation where it can only happen at particular scales.

2492
02:16:39.904 --> 02:16:40.584
Particular scales.

2493
02:16:40.684 --> 02:16:42.484
At very small infinitesimal scales.

2494
02:16:42.824 --> 02:16:45.165
This is why, again, it's like a mathematical problem, right?

2495
02:16:45.205 --> 02:16:45.505
It's not...

2496
02:16:47.014 --> 02:16:50.855
It's infinitesimally small scales, and it's happening only at particular places.

2497
02:16:50.936 --> 02:16:54.157
It's not like a continuous singularity, right?

2498
02:16:54.317 --> 02:16:55.457
It's at the center of a black hole.

2499
02:16:55.477 --> 02:16:56.218
It's a single spot.

2500
02:16:56.778 --> 02:16:59.339
Okay, so that's them saying, okay, partial regularity.

2501
02:16:59.359 --> 02:17:10.483
In other words, you've got these decades of work that is placing a very tight fence around any possible singularity, but we haven't proved that that fence has nothing in it.

2502
02:17:11.487 --> 02:17:12.971
There could still be a singularity.

2503
02:17:13.312 --> 02:17:15.177
It just has to be constrained to these things.

2504
02:17:15.397 --> 02:17:18.486
And so the box in which...

2505
02:17:19.633 --> 02:17:21.134
The convective term wins.

2506
02:17:22.194 --> 02:17:23.874
We've now created a very small box.

2507
02:17:23.934 --> 02:17:24.135
Yes.

2508
02:17:24.315 --> 02:17:24.895
It can still win.

2509
02:17:25.015 --> 02:17:25.575
It could still win.

2510
02:17:26.035 --> 02:17:27.776
Against viscosity, but it's a very small box.

2511
02:17:27.796 --> 02:17:28.856
Very small box, right?

2512
02:17:29.596 --> 02:17:31.737
And this brings up to 2000.

2513
02:17:32.117 --> 02:17:37.779
This is when the Clay Mathematics Institute announces the Millennium Prize problems in mathematics.

2514
02:17:37.839 --> 02:17:43.501
These are seven very famous mathematics problems that were formerly Hilbert's problems.

2515
02:17:44.333 --> 02:17:51.501
Um, that, you know, a bunch of them got solved over in the 20th century, some by John Nash, of course, um, and others.

2516
02:17:51.862 --> 02:17:57.748
And finally, we have, um, seven out of those that were picked as like the big ones.

2517
02:17:58.129 --> 02:18:01.853
The point car rate conjecture has been solved already by Gregory Perlman in 2003.

2518
02:18:01.873 --> 02:18:01.933
Um,

2519
02:18:03.494 --> 02:18:04.835
And that was huge.

2520
02:18:04.895 --> 02:18:09.937
He never accepted the million dollars because he said that someone else deserved it.

2521
02:18:10.277 --> 02:18:11.998
He's kind of like this weird guy.

2522
02:18:12.198 --> 02:18:12.978
I'm going to be honest.

2523
02:18:12.998 --> 02:18:20.321
He's a weird Russian guy, brilliant mathematician, but like lives kind of like in his mom's basement kind of thing.

2524
02:18:20.341 --> 02:18:20.601
Yeah.

2525
02:18:20.721 --> 02:18:22.121
Anyways, brilliant mathematician.

2526
02:18:23.142 --> 02:18:24.462
So here's the problem statement.

2527
02:18:24.682 --> 02:18:26.202
And this is a problem statement by Charles.

2528
02:18:26.282 --> 02:18:35.224
I want to make one quick note here, which is for those viewing who can see the list, I want you to remember this list, right?

2529
02:18:36.084 --> 02:18:42.986
Because this list is going to come into relevance when we talk about the race between the frontier labs.

2530
02:18:43.666 --> 02:18:46.827
And this is the gold medal that they're all chasing after.

2531
02:18:46.887 --> 02:18:47.107
Yeah.

2532
02:18:47.527 --> 02:19:08.222
is and the infinity stone and these are the infinity stones and they are coming after all of these maybe not yang mills and maybe not p versus mp yeah but all the other ones all the other ones reman hypothesis they came after it um or at least something that is related related um navier stokes is the one that we're at right now um yang mills i mean we we did yang mills

2533
02:19:08.722 --> 02:19:09.763
for an episode last year.

2534
02:19:09.783 --> 02:19:10.945
That's one of my favorite episodes.

2535
02:19:10.985 --> 02:19:12.026
Check it out if you haven't.

2536
02:19:12.506 --> 02:19:14.228
And so here is the problem statement.

2537
02:19:14.709 --> 02:19:17.892
So each of these problems have to be well-defined, right?

2538
02:19:17.932 --> 02:19:23.057
Like when the Clay Mathematics Institute is like, this is the problem, the problem has to be well-defined mathematically.

2539
02:19:23.117 --> 02:19:25.880
So Charles Fefferman, field medalist,

2540
02:19:27.021 --> 02:19:36.488
Also a student of Elias Stein at Princeton University, the guy who wrote the complex analysis textbook and my professor of complex analysis also.

2541
02:19:36.789 --> 02:19:40.851
So he's a brother, I guess, an academic brother of Terence Tao.

2542
02:19:41.212 --> 02:19:44.474
Because they both had the same PhD advisor, also field medalist.

2543
02:19:44.834 --> 02:19:53.381
I just want to note for those who might not know, field medalist, meaning that they've won this, the most prestigious mathematics award that is available.

2544
02:19:53.661 --> 02:19:54.562
For those who are under 40.

2545
02:19:55.362 --> 02:19:56.403
For those who are under 40.

2546
02:19:56.423 --> 02:19:56.663
Yes.

2547
02:19:57.084 --> 02:20:01.148
I would say the most prestigious mathematics award that people don't really know about is the Apple Prize.

2548
02:20:01.248 --> 02:20:02.409
We talked about this.

2549
02:20:02.449 --> 02:20:04.871
Which is, I think, like the Nobel Prize of mathematics.

2550
02:20:04.931 --> 02:20:05.192
Fair enough.

2551
02:20:05.212 --> 02:20:06.593
People keep saying it's the Fields Medal.

2552
02:20:06.633 --> 02:20:07.414
I really don't think it is.

2553
02:20:08.575 --> 02:20:10.657
Argue about it in the comments.

2554
02:20:10.937 --> 02:20:16.542
So Charles Fefferman puts out this statement of what it is that we are trying to prove, okay?

2555
02:20:16.562 --> 02:20:16.783
Okay.

2556
02:20:18.232 --> 02:20:22.475
We want the existence and smoothness of Navier-Stokes on R3.

2557
02:20:22.555 --> 02:20:24.076
That means in three dimensions.

2558
02:20:24.216 --> 02:20:25.677
Two dimensions, we've already figured it out.

2559
02:20:25.777 --> 02:20:27.478
Olga said, it's smooth.

2560
02:20:27.598 --> 02:20:28.139
We're good to go.

2561
02:20:28.679 --> 02:20:28.919
Okay.

2562
02:20:29.539 --> 02:20:37.225
A and B say existence and smoothness of Navier-Stokes in three dimensions.

2563
02:20:38.315 --> 02:20:43.399
C and D say the breakdown of Navier-Stokes in three dimensions.

2564
02:20:44.180 --> 02:20:47.442
And just to quickly touch on this.

2565
02:20:47.763 --> 02:20:51.045
So we're saying A and B means viscosity wins.

2566
02:20:51.345 --> 02:20:51.586
Yes.

2567
02:20:51.766 --> 02:20:54.248
And C and D means the convective term wins.

2568
02:20:54.368 --> 02:20:54.848
Exactly.

2569
02:20:55.108 --> 02:20:55.369
Yes.

2570
02:20:55.629 --> 02:20:55.909
Okay.

2571
02:20:55.929 --> 02:20:57.991
So A and B say it's always going to be smooth.

2572
02:20:58.871 --> 02:21:00.112
C and D say it's...

2573
02:21:01.153 --> 02:21:02.394
going to lose.

2574
02:21:02.634 --> 02:21:10.640
In the small fence that we've already defined, the convective term ultimately wins and we get a very discrete small singularity somewhere.

2575
02:21:10.660 --> 02:21:11.041
Exactly.

2576
02:21:11.361 --> 02:21:19.987
And so in A and B, that's a proof of global regularity, and that would mean you need to identify some kind of mechanism that prevents this infinite concentration at

2577
02:21:20.848 --> 02:21:22.149
these very small scales, right?

2578
02:21:22.470 --> 02:21:29.277
In practical terms, one would like some kind of quantity or some kind of scale critical quantity that's controlled for all time.

2579
02:21:29.737 --> 02:21:37.185
And it's some structural argument saying that whatever vortex stretching is happening that we were talking about, it's going to lose out to viscosity, okay?

2580
02:21:38.125 --> 02:21:39.847
A proof for the breakdown would be the opposite.

2581
02:21:39.867 --> 02:21:43.009
You just have to construct a genuine Navier-Stokes flow.

2582
02:21:43.029 --> 02:21:48.474
You have to construct some kind of initial condition in which the concentration reinforces itself.

2583
02:21:48.854 --> 02:21:51.957
That convective term keeps making smaller and smaller things.

2584
02:21:52.618 --> 02:22:01.986
And all of that pressure and nonlinear cancellations, after all of that, the mathematics becomes demonstrably something like a singularity.

2585
02:22:02.006 --> 02:22:02.527
Okay.

2586
02:22:04.828 --> 02:22:18.216
To make an analogy, because this is sort of how people always frame these things when they talk about AI's progress as it relates to these things, which a year ago people didn't think it could compete in the Math Olympiad and now it's solving C&amp;D of a millennium problem.

2587
02:22:18.256 --> 02:22:26.720
But putting that aside, it almost seems, based on this construction we've just described, that it's kind of akin to a counterexample.

2588
02:22:26.840 --> 02:22:28.341
Yes, 100%.

2589
02:22:28.441 --> 02:22:30.543
Versus sort of the global solution.

2590
02:22:30.563 --> 02:22:30.963
Mm-hmm.

2591
02:22:31.683 --> 02:22:34.426
Not minimizing in any way because it's still a big deal.

2592
02:22:34.466 --> 02:22:38.370
But people will nuance and nitpick on that point.

2593
02:22:38.410 --> 02:22:39.130
It's like, great.

2594
02:22:39.150 --> 02:22:40.311
You found one.

2595
02:22:40.972 --> 02:22:47.478
You're looking at this large search space and you found something that has blow up in finite time.

2596
02:22:47.938 --> 02:22:49.860
Well, it's an incredibly small search space.

2597
02:22:49.880 --> 02:22:49.980
Yeah.

2598
02:22:50.781 --> 02:22:51.001
Right?

2599
02:22:51.462 --> 02:22:52.503
And everyone's been looking.

2600
02:22:52.523 --> 02:22:53.323
Looking at it.

2601
02:22:53.363 --> 02:22:53.503
Right?

2602
02:22:53.844 --> 02:22:56.226
But I think you raise a very big point.

2603
02:22:56.326 --> 02:22:59.849
And I think we're going to come to that in the next few examples that I'm showing you.

2604
02:22:59.869 --> 02:23:05.774
Because in 2000, this is when the Clay Mathematics Institute came up with it, right?

2605
02:23:06.334 --> 02:23:12.380
In 2016, big players are involving themselves in this.

2606
02:23:12.900 --> 02:23:14.782
Because they smell the blood in the water.

2607
02:23:15.382 --> 02:23:17.924
That there is a counterexample hanging out somewhere.

2608
02:23:18.385 --> 02:23:18.665
Okay?

2609
02:23:20.093 --> 02:23:26.455
More and more analysis is going into showing that perhaps there are situations where you get singularities.

2610
02:23:26.755 --> 02:23:33.458
So Terence Tao, he creates this averaged version of the Navier-Stokes equations.

2611
02:23:33.758 --> 02:23:39.800
And he proved that the standard energy conservation laws are mathematically too weak to prevent a singularity.

2612
02:23:40.420 --> 02:23:44.202
So he created this like averaged version of the Navier-Stokes equations.

2613
02:23:44.222 --> 02:23:45.662
It's kind of like a blurred out version.

2614
02:23:45.682 --> 02:23:46.682
Maybe you can think about that.

2615
02:23:46.722 --> 02:23:48.223
It's like a stunt double equation.

2616
02:23:48.720 --> 02:23:49.640
of the fluid equations.

2617
02:23:50.081 --> 02:23:55.623
And from the outside, this thing obeys all of the exact same conservation laws that the real Navier-Stokes equations does.

2618
02:23:56.223 --> 02:24:02.946
But tau proved that this stunt double definitively explodes in finite time.

2619
02:24:03.507 --> 02:24:03.767
Okay?

2620
02:24:04.027 --> 02:24:12.951
So this averaged version can create singularities, which means that the principles that you would have used to prove that it wasn't possible are...

2621
02:24:13.932 --> 02:24:17.514
are also not going to work for the real thing.

2622
02:24:18.034 --> 02:24:21.816
If it was global, you wouldn't be able to have an average stunt double that blew up.

2623
02:24:21.896 --> 02:24:22.897
Yes, exactly.

2624
02:24:23.177 --> 02:24:31.041
So there's a chance now that at least the original arguments that people were using, saying, oh, energy conservation and things like that, that might not work anymore.

2625
02:24:31.061 --> 02:24:31.521
Okay?

2626
02:24:31.942 --> 02:24:34.383
So A and B are already not looking great.

2627
02:24:34.663 --> 02:24:34.863
Right?

2628
02:24:34.963 --> 02:24:35.684
Maybe there is...

2629
02:24:36.804 --> 02:24:38.505
A singularity that's lurking somewhere.

2630
02:24:39.065 --> 02:24:39.305
Okay?

2631
02:24:39.725 --> 02:24:49.730
And for decades, mathematicians tried to trigger this blow-up using something like a fractal shape, like a whirlpool that's shrinking into smaller and smaller whirlpools and so on.

2632
02:24:50.710 --> 02:24:51.671
This usually failed.

2633
02:24:52.571 --> 02:24:54.693
Because it required an infinite amount of energy.

2634
02:24:54.813 --> 02:24:56.494
And crucially, you can't do that.

2635
02:24:56.774 --> 02:25:02.199
One of the things that Pfefferman said is, you can't just stir it infinitely fast and be like, oh, I got infinity.

2636
02:25:02.219 --> 02:25:05.401
That's not the interesting question we're trying to ask.

2637
02:25:05.501 --> 02:25:10.065
Okay, so you supplied an infinite amount of energy and you got an infinite amount of velocity.

2638
02:25:10.145 --> 02:25:10.786
Great.

2639
02:25:11.106 --> 02:25:19.593
No, it's supplying a finite amount of energy to get an infinite amount of velocity because the nonlinear convective term is concentrating energy.

2640
02:25:20.313 --> 02:25:22.015
into smaller and smaller structures.

2641
02:25:22.215 --> 02:25:27.000
So both finite energy and finite time are both relevant in the context of this.

2642
02:25:27.020 --> 02:25:27.380
Yes, 100%.

2643
02:25:27.861 --> 02:25:35.088
And so Diego Cordova and Luis Martinez Zorora and Fan Zhang in 2023 and 2025, this is very recent,

2644
02:25:37.190 --> 02:25:43.613
They changed the paradigm by abandoning this whole single shape thing and relying on something called a layer cascade.

2645
02:25:43.993 --> 02:25:51.756
It's kind of like a bunch of nested gears or, you know, those Russian dolls where you have like a big doll and then smaller doll and smaller doll and smaller dolls.

2646
02:25:52.116 --> 02:25:53.236
That's what they were constructing.

2647
02:25:53.537 --> 02:26:01.740
Instead of a single vortex that they do this vortex stretching in, they've got slow moving outer vortex and then a medium inner vortex.

2648
02:26:02.320 --> 02:26:06.243
and then a smaller inner vortex that spins faster because of the medium vortex.

2649
02:26:06.523 --> 02:26:13.009
And the medium vortex is kind of like a parlay, or like, not a parlay, like an intermediary between the outer and the inner.

2650
02:26:13.949 --> 02:26:17.952
And it acts upon an even smaller layer, and so on and so forth.

2651
02:26:18.013 --> 02:26:24.878
And you can design this geometry so that the feedback between these gears doesn't disrupt the big gears.

2652
02:26:25.438 --> 02:26:27.200
And the feedback between the smaller gears...

2653
02:26:28.622 --> 02:26:31.504
goes up, up, and up, and you can get unbounded velocity.

2654
02:26:32.344 --> 02:26:32.624
Okay?

2655
02:26:33.365 --> 02:26:38.428
And if you have an infinite number of gears, then perhaps you can have an infinite velocity.

2656
02:26:38.728 --> 02:26:42.790
And so the mechanism they called a dynamical amplification across scales.

2657
02:26:43.170 --> 02:26:48.794
And they used this to prove finite time blowup for the unforced 3D Euler equations.

2658
02:26:49.194 --> 02:26:51.435
Remember the Euler equations without viscosity?

2659
02:26:51.455 --> 02:26:52.336
Without viscosity.

2660
02:26:52.356 --> 02:26:52.456
Yeah.

2661
02:26:52.816 --> 02:26:56.739
They showed that without the smoothing term, you could actually have finite blow up.

2662
02:26:56.919 --> 02:27:00.761
The pressure won't do as much because the pressure also does a little bit of smoothing.

2663
02:27:00.941 --> 02:27:05.644
Because it was both the pressure and the external force were the only two of the three on the force part of the equation.

2664
02:27:05.665 --> 02:27:05.865
Yeah.

2665
02:27:06.205 --> 02:27:07.486
And so this is unforced.

2666
02:27:07.506 --> 02:27:08.246
So no force.

2667
02:27:08.707 --> 02:27:09.927
You just set up an initial.

2668
02:27:09.967 --> 02:27:10.588
Ah, okay.

2669
02:27:10.608 --> 02:27:10.868
Okay.

2670
02:27:11.088 --> 02:27:11.328
Yeah.

2671
02:27:11.488 --> 02:27:11.688
No force.

2672
02:27:11.708 --> 02:27:12.549
There's no stirring.

2673
02:27:12.929 --> 02:27:14.150
There's no force.

2674
02:27:14.170 --> 02:27:21.735
You just set up an initial condition, and the scales converge in such a way that you get an infinite blowup.

2675
02:27:21.775 --> 02:27:22.495
Now, this is huge.

2676
02:27:22.795 --> 02:27:26.958
They're chipping away at each of the variables on the right side of the equation.

2677
02:27:27.338 --> 02:27:27.578
Yes.

2678
02:27:27.738 --> 02:27:28.199
Wow.

2679
02:27:30.232 --> 02:27:34.853
The problem was, so that was for the unforced 3D Euler equations, right?

2680
02:27:35.133 --> 02:27:37.733
But obviously we're interested in the ones that have viscosity.

2681
02:27:38.253 --> 02:27:44.435
So there's this one type of equation called the hypodissipative Navier-Stokes equations.

2682
02:27:44.555 --> 02:27:52.016
It's a variant that utilizes this thing called a fraction Laplacian instead of the actual Laplacian.

2683
02:27:52.096 --> 02:27:57.837
The del squared u that we had for viscosity, the triangle squared u, that's a Laplacian.

2684
02:27:58.357 --> 02:28:00.539
This thing has a fractional Laplacian.

2685
02:28:00.559 --> 02:28:01.759
Fractional Laplacian.

2686
02:28:01.779 --> 02:28:06.843
And they showed that in this case, we can use the same sort of paradigm and get blow up.

2687
02:28:08.043 --> 02:28:14.388
The only problem was that they required a forcing term that was rough.

2688
02:28:15.224 --> 02:28:16.285
Meaning it's not smooth.

2689
02:28:16.726 --> 02:28:20.751
Like, the forcing term itself has tiny little bits of infinity.

2690
02:28:21.553 --> 02:28:25.898
Like, the way you're stirring it has, like, little jolts of infinite stuff.

2691
02:28:25.938 --> 02:28:28.181
You're pumping infinity into your stirring.

2692
02:28:28.742 --> 02:28:29.183
And so...

2693
02:28:30.817 --> 02:28:39.791
We're getting closer, though, because the infinity is small enough that you can maybe figure out a way to smooth it and still preserve the dynamics.

2694
02:28:40.131 --> 02:28:45.339
And so I just want to pause really quick because what we're sort of saying is human mathematicians...

2695
02:28:45.980 --> 02:29:05.437
have now taken this Navier Stokes Millennium Prize, looking at finite blow-up, and then they started with looking at taking out the smoothness and taking out all the force.

2696
02:29:05.657 --> 02:29:09.881
We got some solutions, which are effectively the Euler equations without the F.

2697
02:29:11.022 --> 02:29:15.526
And then we've now started adding in, back in the viscosity.

2698
02:29:15.726 --> 02:29:16.987
But that weird version.

2699
02:29:17.007 --> 02:29:18.929
But the viscosity wasn't smooth.

2700
02:29:19.269 --> 02:29:19.729
It was rough.

2701
02:29:20.430 --> 02:29:38.464
And so when people see this smoothness and roughness, part of what we're saying is the viscosity term has some special sauce that doesn't quite get you to the natural smoothness that arises in the equations with viscosity un-

2702
02:29:39.285 --> 02:29:40.807
Yeah, the raw viscosity.

2703
02:29:40.827 --> 02:29:43.530
The raw viscosity.

2704
02:29:43.610 --> 02:29:46.793
But it's still important because these are fundamental.

2705
02:29:46.933 --> 02:29:50.237
No, and this is what even OpenAI's thing is built on.

2706
02:29:50.718 --> 02:29:50.978
Okay?

2707
02:29:51.258 --> 02:29:56.864
It's like, there's a reason why the Navier Stokes is the first of the six infinity stones to fall.

2708
02:29:56.944 --> 02:29:57.545
Which we just talked about.

2709
02:29:57.565 --> 02:30:00.188
It's because we got so close to...

2710
02:30:00.348 --> 02:30:12.163
as human beings it's like a racetrack and someone was on the lap the fourth lap of whatever the 800 meter yeah and uh you know open eye just shows up grabs a baton and finishes yeah yeah yeah

2711
02:30:13.397 --> 02:30:14.178
Not minimizing.

2712
02:30:14.218 --> 02:30:15.438
Not minimizing, of course.

2713
02:30:15.819 --> 02:30:19.241
But I'm just saying, a lot of human input has gone into this, right?

2714
02:30:19.281 --> 02:30:25.004
So Cordoba and Martinez-Zorora, they established this layer-by-layer scale amplification.

2715
02:30:25.085 --> 02:30:29.027
It's a huge deal because everyone can sense that we're getting close.

2716
02:30:29.267 --> 02:30:29.547
Okay?

2717
02:30:30.188 --> 02:30:38.193
And shortly before the OpenAI announcement, mathematicians Tristan Buckmaster and Levant Alpoghe, they published preprints that elevated this methodology and,

2718
02:30:38.453 --> 02:30:44.215
to accommodate for smooth forcing of Euler, meaning you've got the Euler equations, but now with a smooth force.

2719
02:30:44.535 --> 02:30:49.957
The Boyzonesque equation, which is something that's like kind of similar to the Navier-Stokes equations.

2720
02:30:51.198 --> 02:31:00.521
And this is where the sort of drama starts, but I'm going to save that for later because a few days after that, OpenAI announces that they have solved it.

2721
02:31:01.760 --> 02:31:02.461
Here we go.

2722
02:31:02.501 --> 02:31:02.901
Here we go.

2723
02:31:03.261 --> 02:31:05.123
On the Navier Stokes Millennium Prize problem.

2724
02:31:05.403 --> 02:31:06.303
They announce it on Twitter.

2725
02:31:06.323 --> 02:31:07.384
They have it on their website.

2726
02:31:07.845 --> 02:31:12.388
And the actual paper is something like 165 pages long.

2727
02:31:13.209 --> 02:31:23.256
The theorem proved by OpenAI constructs a smooth, compactly supported external force that drives a fluid starting from rest into a singularity.

2728
02:31:24.537 --> 02:31:25.218
That's what they're doing.

2729
02:31:25.778 --> 02:31:27.760
They're solving C and D. Right.

2730
02:31:28.160 --> 02:31:28.360
Right.

2731
02:31:28.560 --> 02:31:30.602
They're saying that I have found a way.

2732
02:31:31.443 --> 02:31:39.631
To force a singularity into... We found a situation where the convective term beats the viscosity.

2733
02:31:39.731 --> 02:31:40.172
Exactly.

2734
02:31:40.512 --> 02:31:41.473
Yeah.

2735
02:31:41.913 --> 02:31:47.379
And because the force is smooth and bounded, the singularity is not because the force is just infinite.

2736
02:31:47.839 --> 02:31:48.039
Right?

2737
02:31:48.159 --> 02:31:51.703
It's literally the fluid itself, the velocity field itself...

2738
02:31:52.143 --> 02:31:52.624
doing this.

2739
02:31:52.964 --> 02:31:58.468
The force is trying to maintain some kind of structural tautness of that liquid.

2740
02:31:59.009 --> 02:32:01.611
And the singularity happens at t equals 1.

2741
02:32:02.672 --> 02:32:04.213
It doesn't happen at t equals infinity.

2742
02:32:04.493 --> 02:32:06.555
This is a finite time blow up.

2743
02:32:07.035 --> 02:32:10.758
Okay, so let's get into some of the stuff that I understood from their paper.

2744
02:32:11.299 --> 02:32:19.345
This is unlike the anthropic paper about Riemann-Zeta, because in the anthropic version, they gave us a full account of all of the agents that,

2745
02:32:19.780 --> 02:32:25.743
and like even a transcript of the inner thoughts of these agents and how they coordinated with all these sub-agents.

2746
02:32:26.063 --> 02:32:30.946
So I had this really nice story for our podcast where I was talking about how some of the agents like,

2747
02:32:31.954 --> 02:32:37.517
you know, something crashed and like they had saved a little lemma of their proof on the hard disk.

2748
02:32:38.058 --> 02:32:39.479
Those are the best parts of the story.

2749
02:32:39.499 --> 02:32:43.841
Yeah, it was crazy that we got this like inner version of what the agents were doing.

2750
02:32:44.162 --> 02:32:46.603
Here, OpenAI is not doing that.

2751
02:32:46.943 --> 02:32:54.728
But in any case, we can still kind of try to understand how they constructed the singularity.

2752
02:32:54.748 --> 02:32:56.089
The core of the singularity is...

2753
02:32:56.549 --> 02:33:00.270
in a axisymmetric vortex that's centered at the origin.

2754
02:33:00.791 --> 02:33:02.871
And this is figure one that they show.

2755
02:33:04.192 --> 02:33:07.593
We start out with a vortex that's yay big and yay tall.

2756
02:33:08.593 --> 02:33:16.476
And the point is that as the vortex shrinks, the vortex is going to stretch, right?

2757
02:33:16.976 --> 02:33:22.238
The whole thing is shrinking, but the radius is shrinking faster than the height.

2758
02:33:24.526 --> 02:33:47.748
that's the whole idea because before this is so funny because before we talked about the hurricane going to the tornado and the idea was the height was increasing yeah so fast because of the conservation of angular momentum and all they're just saying into this mathematical construction where you don't need the conservation of angular momentum you still do you still do but they're hacking their way around okay okay okay yeah yeah they're hacking their way around okay okay okay

2759
02:33:48.688 --> 02:33:58.152
They're saying that as the thing shrinks, the radius is shrinking faster than my height, which you can kind of see here in this example, right?

2760
02:33:58.632 --> 02:34:02.213
The blue part, the radius is kind of shrinking faster than the height.

2761
02:34:02.473 --> 02:34:05.575
But as the radius shrinks, it's going to spin faster and faster and faster.

2762
02:34:05.595 --> 02:34:09.356
That's how they're... And I'm going to get an infinite amount of velocity.

2763
02:34:09.656 --> 02:34:10.357
That's okay.

2764
02:34:10.417 --> 02:34:11.017
I get it now.

2765
02:34:11.117 --> 02:34:11.378
Yep.

2766
02:34:11.838 --> 02:34:21.006
And there was a really nice GIF that was prepared by Jayanta Padhikar from the Wolfram community staff picks September 14, 2026.

2767
02:34:21.046 --> 02:34:24.609
You should really check out the website and it shows sort of what's happening.

2768
02:34:25.089 --> 02:34:25.289
Okay.

2769
02:34:25.509 --> 02:34:31.114
In this case, the camera you have to imagine is zooming in because the whole time this entire thing is shrinking.

2770
02:34:31.174 --> 02:34:31.634
Right here.

2771
02:34:31.674 --> 02:34:37.479
It's kind of showing that like, it's kind of like the same size the whole time, but that's because the camera is zooming in with the vortex.

2772
02:34:37.499 --> 02:34:37.579
Yes.

2773
02:34:37.659 --> 02:34:37.920
Yes.

2774
02:34:39.161 --> 02:34:43.123
But here what you're seeing, let's look at this animation in a great bit of detail.

2775
02:34:44.925 --> 02:34:47.687
The radius is going down and down, you can see.

2776
02:34:48.347 --> 02:34:53.891
The axial length is kind of going up, and the aspect ratio is going up.

2777
02:34:54.571 --> 02:35:02.717
That's because the radius is shrinking as the height increases, but it's not one of those PowerPoints where you hold shift and

2778
02:35:03.097 --> 02:35:05.358
And it preserves the shape.

2779
02:35:05.718 --> 02:35:06.959
In this case, it's stretching.

2780
02:35:07.299 --> 02:35:08.760
This is vortex stretching.

2781
02:35:09.060 --> 02:35:13.242
And that inner vortex that you see, you see there's an inner vortex and an outer vortex?

2782
02:35:13.402 --> 02:35:16.844
That inner vortex is the thing that is causing the singularity.

2783
02:35:17.264 --> 02:35:26.328
That inner vortex is the thing that is making stuff move at faster and faster speeds all the way to infinity as you approach time t equals 1.

2784
02:35:26.648 --> 02:35:27.929
That makes total sense.

2785
02:35:28.009 --> 02:35:28.109
Yeah.

2786
02:35:28.489 --> 02:35:44.776
And it's so interesting because, again, we've built this construction where I can actually walk away now looking at this and understand why does what I'm looking at equal finite time blowup of Navier Stokes.

2787
02:35:44.976 --> 02:35:49.497
And it's funny because we talk a lot on the show about

2788
02:35:58.500 --> 02:36:01.064
how we're standing on the shoulders of so much understanding.

2789
02:36:01.925 --> 02:36:07.394
Um, and I'll, I'll pause on this because we'll get to it in a minute, but it's so interesting.

2790
02:36:08.055 --> 02:36:10.118
Um, that, that,

2791
02:36:12.023 --> 02:36:18.269
we are accelerating our ability to build on the learnings of those who came before us.

2792
02:36:18.309 --> 02:36:19.710
Yeah.

2793
02:36:19.850 --> 02:36:20.891
It's really, really cool.

2794
02:36:21.231 --> 02:36:27.677
And there's a few more things that I want to talk about this OpenAI paper that I found as a physicist quite interesting.

2795
02:36:28.077 --> 02:36:35.204
Because one of the key things that I love about physics when I was doing statistical mechanics is scaling arguments.

2796
02:36:35.864 --> 02:36:48.595
Meaning, as I approach criticality, criticality is a very big thing in statistical mechanics when we talk about critical phenomenon, magnetic materials, things like that.

2797
02:36:50.597 --> 02:36:52.659
It's all about the exponents.

2798
02:36:52.979 --> 02:36:58.244
It's about, as I get closer and closer to the critical phenomenon, how does my stuff blow up?

2799
02:36:58.804 --> 02:37:01.207
And how does my stuff approach infinity?

2800
02:37:01.467 --> 02:37:02.067
For example...

2801
02:37:04.017 --> 02:37:04.638
In a magnet.

2802
02:37:04.718 --> 02:37:10.849
If I get close and closer to the Curie temperature, which is where my magnet goes and remains...

2803
02:37:11.329 --> 02:37:16.358
If I have a magnet and I cool it down to a certain temperature, then even when I remove...

2804
02:37:17.197 --> 02:37:37.548
magnetic field on the outside it'll remain a magnet you know how like you take a paper clip and you you rub it it becomes a magnet but then like you leave it it'll stop being a magnet if you cool this paper clip down it's going to remain magnetized because all of the spins are kind of frozen in there's a specific temperature where that happens and as my material approaches that temperature

2805
02:37:38.088 --> 02:37:39.248
stuff starts blowing up.

2806
02:37:39.328 --> 02:37:45.791
In this case, the correlation length will start blowing up, meaning a spin can now talk to a spin that's infinitely far away.

2807
02:37:46.071 --> 02:37:47.531
At least that's what the model suggests.

2808
02:37:48.092 --> 02:37:55.174
And some of the stuff in this paper reminded me of that class that I was taking.

2809
02:37:55.314 --> 02:37:57.035
Again, that was another class with Robin Bruinsway.

2810
02:37:57.055 --> 02:37:58.635
It was a different class than fluid mechanics.

2811
02:37:58.956 --> 02:38:00.696
But in any case, it was one of my favorite courses.

2812
02:38:01.496 --> 02:38:05.418
So let's take a look at some of these scaling arguments that the OpenAI paper is claiming.

2813
02:38:06.078 --> 02:38:19.047
So one thing that's very cool is they recast the time to instead of being just time as it's approaching one, we talk about tau, which is a countdown time as I approach one.

2814
02:38:19.548 --> 02:38:24.031
So, you know, normal clock time would be like zero seconds, 0.5 seconds, 0.7 seconds approaching one.

2815
02:38:26.417 --> 02:38:29.079
But imagine taking one minus that time.

2816
02:38:29.579 --> 02:38:34.522
Then I'm like one second away from singularity, 0.5 seconds away from singularity.

2817
02:38:34.602 --> 02:38:36.503
I'm counting down to zero.

2818
02:38:36.763 --> 02:38:38.904
And zero means I've gotten to singularity.

2819
02:38:39.204 --> 02:38:41.926
So first thing they do is they recast time to become tau.

2820
02:38:42.226 --> 02:38:44.107
And tau is like a countdown.

2821
02:38:44.587 --> 02:38:47.229
It's like a ticking thing that is going down to zero.

2822
02:38:47.249 --> 02:38:47.649
Mm-hmm.

2823
02:38:48.138 --> 02:38:53.462
And they say, what is happening to my radial and my length scales as I approach zero?

2824
02:38:54.502 --> 02:38:57.524
For one, the radial scale goes down like the square root of tau.

2825
02:38:58.085 --> 02:38:59.546
Okay, so it's still going to zero, right?

2826
02:38:59.566 --> 02:39:03.628
Because square root of zero is still, when tau goes to zero, the radial scale is going to zero.

2827
02:39:03.969 --> 02:39:05.650
But it's going like the square root.

2828
02:39:05.710 --> 02:39:11.814
On the other hand, the axial scale is going like tau to the one-half minus h, where h is some non-zero number.

2829
02:39:19.458 --> 02:39:24.100
I still have a little bit of stuff left.

2830
02:39:24.280 --> 02:39:24.700
Okay?

2831
02:39:25.381 --> 02:39:27.622
The aspect ratio is actually going to infinity.

2832
02:39:29.742 --> 02:39:33.764
As I squish this thing, this guy is not shrinking that fast.

2833
02:39:34.244 --> 02:39:41.447
And so my height versus my radius, if I were to divide the two, that aspect ratio is actually going to infinity.

2834
02:39:41.608 --> 02:39:42.448
Mm-hmm.

2835
02:39:42.468 --> 02:39:42.608
Mm-hmm.

2836
02:39:43.238 --> 02:39:44.540
Which is key.

2837
02:39:44.760 --> 02:39:45.060
Okay.

2838
02:39:45.080 --> 02:39:54.012
Because if the aspect ratio is going to infinity, which in this case, what we're saying is that aspect ratio is the vortex is becoming infinitely slender.

2839
02:39:54.252 --> 02:39:54.552
Mm-hmm.

2840
02:39:55.374 --> 02:39:55.654
Yes.

2841
02:39:56.074 --> 02:39:57.076
And so what's happening is...

2842
02:39:58.672 --> 02:40:00.313
The vortex is going like this.

2843
02:40:00.873 --> 02:40:02.093
It's getting smaller and smaller.

2844
02:40:02.333 --> 02:40:06.114
But because of incompressibility, the fluid has to go somewhere.

2845
02:40:06.434 --> 02:40:07.735
It's going to start shooting out.

2846
02:40:08.315 --> 02:40:12.996
And the smaller this thing gets and the taller this thing gets, the faster it's going to shoot out.

2847
02:40:13.677 --> 02:40:24.080
And so my azimuthal velocity is going to go to infinity because it's going to shoot out of this vortex increasingly fast as I shrink this thing down and down and down.

2848
02:40:24.100 --> 02:40:24.480
Mm-hmm.

2849
02:40:25.703 --> 02:40:28.525
And this shows the characteristic velocity magnitudes.

2850
02:40:30.227 --> 02:40:41.297
Again, you've got these scaling arguments where stuff goes like the square root, but then one of them goes like the square root minus h, and the h is not zero.

2851
02:40:41.317 --> 02:40:46.802
The h is some small number, but because it's negative, that means it's in the reciprocal.

2852
02:40:46.842 --> 02:40:48.024
So when t goes to zero...

2853
02:40:49.053 --> 02:40:51.195
One over zero, you go to infinity.

2854
02:40:51.815 --> 02:40:52.016
Right?

2855
02:40:52.236 --> 02:40:57.420
So the azimuthal velocities, the shooting up and down is going to infinity.

2856
02:40:58.061 --> 02:40:58.261
Right?

2857
02:40:58.381 --> 02:41:01.384
Even though the radial velocity is on the order of going to zero.

2858
02:41:02.284 --> 02:41:03.045
Actually, no.

2859
02:41:03.085 --> 02:41:04.446
In this case, it's also going to infinity.

2860
02:41:05.067 --> 02:41:05.707
Okay.

2861
02:41:05.827 --> 02:41:06.948
And it's...

2862
02:41:07.529 --> 02:41:07.949
I'll pause.

2863
02:41:08.129 --> 02:41:08.930
Let's continue.

2864
02:41:09.050 --> 02:41:12.093
The next thing I want to show is, is there infinite kinetic energy?

2865
02:41:12.153 --> 02:41:13.254
Because if there is, then we're...

2866
02:41:14.512 --> 02:41:15.172
then that's not good.

2867
02:41:15.252 --> 02:41:15.933
It doesn't, yeah.

2868
02:41:15.993 --> 02:41:16.653
That doesn't count, then.

2869
02:41:16.673 --> 02:41:17.153
It doesn't count.

2870
02:41:17.293 --> 02:41:18.473
Well, let's look at it.

2871
02:41:18.733 --> 02:41:24.856
The kinetic energy is determined by the spatial integral of the velocity squared, right?

2872
02:41:24.876 --> 02:41:25.976
Because 1.5 mv squared.

2873
02:41:26.016 --> 02:41:26.136
Yep.

2874
02:41:26.316 --> 02:41:27.557
So it's just the velocity squared.

2875
02:41:27.797 --> 02:41:30.878
Well, velocity squared goes like tau to the minus 1 minus 2h.

2876
02:41:32.143 --> 02:41:45.450
And the volume, because it's the radius multiplied by the height, radius squared multiplied by the height, the volume goes like tau to the 3 halves minus h. You can just see the previous slides that I showed you and calculate the volume.

2877
02:41:45.710 --> 02:41:51.073
You multiply the velocity squared by the volume, and you get something that goes like tau to the 1 half minus 3h.

2878
02:41:51.434 --> 02:41:54.435
When tau goes to zero, that thing goes to zero.

2879
02:41:54.515 --> 02:41:54.815
Yeah.

2880
02:41:55.536 --> 02:41:57.437
So the energy is going to zero, right?

2881
02:41:58.454 --> 02:42:00.575
But the velocity is going to infinity.

2882
02:42:01.056 --> 02:42:01.936
It's not an infinite.

2883
02:42:01.956 --> 02:42:03.297
They're not putting infinite kinetic energy.

2884
02:42:03.317 --> 02:42:05.838
They're not putting infinite kinetic energy into the system.

2885
02:42:06.178 --> 02:42:07.619
It is a runaway singularity.

2886
02:42:07.659 --> 02:42:07.939
Yes.

2887
02:42:08.140 --> 02:42:11.081
That's not correlated to the amount of energy in the system.

2888
02:42:11.101 --> 02:42:11.341
Yes.

2889
02:42:12.822 --> 02:42:15.364
The energy in the system is not causing the singularity.

2890
02:42:15.504 --> 02:42:18.145
The singularity is being caused by the fluid itself.

2891
02:42:18.165 --> 02:42:19.466
The fluid dynamics itself.

2892
02:42:19.486 --> 02:42:19.666
Yeah.

2893
02:42:19.966 --> 02:42:24.229
The way that they've constructed this geometry and this tiny bit of forcing that they have.

2894
02:42:24.509 --> 02:42:25.389
Let's look at the forcing.

2895
02:42:25.409 --> 02:42:26.210
So that's figure two.

2896
02:42:27.283 --> 02:42:31.444
Here you're looking at a bird's eye, like sort of down the tornado hole of the thing.

2897
02:42:31.744 --> 02:42:32.525
And you zoom in.

2898
02:42:32.825 --> 02:42:37.866
The way they're forcing this thing is by pushing out and pushing in simultaneously.

2899
02:42:38.346 --> 02:42:38.646
Okay?

2900
02:42:38.846 --> 02:42:41.887
So you push out to the left and you push in to the right.

2901
02:42:42.267 --> 02:42:48.829
What that does is impart the same angular momentum, but it cancels out kind of the radial thing.

2902
02:42:49.249 --> 02:42:52.130
So all you're doing is turning this thing in some sense.

2903
02:42:52.170 --> 02:42:53.531
And not making it wider?

2904
02:42:53.551 --> 02:42:54.151
Yeah, yeah.

2905
02:42:55.168 --> 02:42:55.728
Oh, that's nice.

2906
02:42:55.848 --> 02:42:56.489
It's kind of nice.

2907
02:42:56.569 --> 02:42:57.089
That's nice.

2908
02:42:57.149 --> 02:42:57.369
Yeah.

2909
02:42:57.510 --> 02:42:59.671
And also the pushes are localized.

2910
02:42:59.751 --> 02:43:00.111
Yeah.

2911
02:43:00.131 --> 02:43:01.532
They're not in a total ring.

2912
02:43:01.832 --> 02:43:05.755
Like if you zoom in on that part of the ring, they're localized to these little tiny spots.

2913
02:43:06.055 --> 02:43:10.098
This goes back to the convective term versus the vector field.

2914
02:43:10.138 --> 02:43:13.720
Like the changes are happening at a very small part.

2915
02:43:14.000 --> 02:43:21.365
And then those changes are then ricocheting into the smaller and smaller vortices to create that thing.

2916
02:43:21.725 --> 02:43:22.086
Right?

2917
02:43:22.126 --> 02:43:22.646
This is good.

2918
02:43:22.886 --> 02:43:24.286
I mean, it's very cool.

2919
02:43:24.306 --> 02:43:37.649
And this is on the backs of mathematicians over the ages, but I still think this definitively shows that there are situations where the convective term wins no matter how hard the viscosity tries.

2920
02:43:38.489 --> 02:43:45.051
The vendetta was not successful by V. And so you would think that this is the end of Navier-Stokes.

2921
02:43:45.331 --> 02:43:45.811
Not quite.

2922
02:43:46.251 --> 02:43:49.672
It is the end of the Millennium Problem because the Clay Mathematics Institute

2923
02:43:50.793 --> 02:43:55.820
Charles Fefferman had a very stringent criteria, and this thing meets that stringent criteria.

2924
02:43:55.880 --> 02:43:57.162
It's been verified by Lean.

2925
02:43:57.502 --> 02:44:00.226
Crucially, that does not mean that it is true, right?

2926
02:44:00.567 --> 02:44:06.235
Lean could have errors, so we're still waiting, and the Clay Mathematics Institute actually is also waiting for two years.

2927
02:44:06.695 --> 02:44:08.196
Before it says, yep, this is done.

2928
02:44:08.556 --> 02:44:15.901
So there needs to be human mathematicians that actually understand this 165-page proof and tell us, yep, we're good to go.

2929
02:44:16.822 --> 02:44:25.127
But the Navier-Stokes problem itself is actually also not fully put to rest because they have solved the forced version, right?

2930
02:44:25.167 --> 02:44:26.788
This required a little bit of paddling.

2931
02:44:27.228 --> 02:44:32.053
In some sense, you took an oar and you like paddled the vortex to become what it is.

2932
02:44:33.214 --> 02:44:35.076
What about the unforced scenario?

2933
02:44:35.436 --> 02:44:35.636
Right.

2934
02:44:35.957 --> 02:44:37.318
That is still an open question.

2935
02:44:37.538 --> 02:44:44.585
And Princeton mathematician Stan Palasek actually identified a problem with this particular problem.

2936
02:44:45.868 --> 02:44:50.190
proof that would make it impossible to apply to an unforced scenario.

2937
02:44:50.690 --> 02:44:53.371
So like this geometry is not going to work for the unforced.

2938
02:44:53.391 --> 02:45:00.114
This only works because you have those little paddling terms that are imparting the finite amount of energy into that cascade.

2939
02:45:00.955 --> 02:45:05.817
So without that hand of God forcing term, it doesn't work.

2940
02:45:05.837 --> 02:45:08.918
And so that part at least is still...

2941
02:45:10.225 --> 02:45:11.005
An open question.

2942
02:45:11.045 --> 02:45:12.926
This natural, unforced setting.

2943
02:45:13.446 --> 02:45:24.891
And this is, as I've seen some of the dialogue about this, this is where, you know, so Scientific American had an article about this and they were like, you know, OpenAI did not solve Navier Stokes.

2944
02:45:25.711 --> 02:45:30.853
And, you know, what they were sort of trying to argue is they didn't solve A and B.

2945
02:45:31.574 --> 02:45:32.554
They solved C and D.

2946
02:45:32.874 --> 02:45:38.195
Like, that's the argument that they were making, but they made it seem like it was complicated.

2947
02:45:38.235 --> 02:45:40.676
No, but the Clay Mathematics Institute said any of the four.

2948
02:45:41.396 --> 02:45:41.896
Correct.

2949
02:45:42.136 --> 02:45:53.799
And this is where, again, I'm just trying to talk about the nuance of what actually was done versus sometimes the way in which it gets then presented in media downstream.

2950
02:45:53.979 --> 02:45:54.219
Yeah.

2951
02:45:54.479 --> 02:45:54.639
And –

2952
02:45:55.679 --> 02:46:01.002
You know, some people now are arguing, well, the way it's defined by clay mathematics is the problem.

2953
02:46:01.262 --> 02:46:01.502
Yeah.

2954
02:46:01.622 --> 02:46:03.483
And it's like, okay, you can make that argument.

2955
02:46:03.503 --> 02:46:04.864
You weren't arguing that before.

2956
02:46:04.884 --> 02:46:05.124
Right.

2957
02:46:05.364 --> 02:46:17.491
But why are we, it's funny that now that it starts when it's been solved by something that people, that either makes people feel uncomfortable or they feel like it's by an entity that didn't give credit to those before them or whatever it might be.

2958
02:46:26.442 --> 02:46:31.883
knowing we were going to go through this, I focused on the larger environment.

2959
02:46:32.523 --> 02:46:52.068
And now having walked through this, with nuances applied, six months ago or a year ago, if you said that a model, even given all of the human advancement, had figured out finite time blow up for Navier Stokes...

2960
02:46:52.258 --> 02:46:53.419
People would have laughed in your face.

2961
02:46:53.519 --> 02:46:53.739
Yeah.

2962
02:46:54.160 --> 02:46:55.081
They would have laughed at you.

2963
02:46:55.201 --> 02:46:55.381
Yeah.

2964
02:46:55.721 --> 02:47:00.126
I mean, those are the guys who are like, yeah, wake me up when AI solves a millennium problem.

2965
02:47:00.306 --> 02:47:02.088
And now the millennium problems are ill-defined.

2966
02:47:02.228 --> 02:47:02.468
Yeah.

2967
02:47:02.508 --> 02:47:03.389
You know, I don't know.

2968
02:47:04.390 --> 02:47:06.472
I don't know.

2969
02:47:06.832 --> 02:47:09.315
So I want to let people know here.

2970
02:47:09.855 --> 02:47:31.627
This is the best explanation of Navier Stokes in the context of OpenAI's announcement and the history of this subject at a level that is accessible to most general people but is not meant to necessarily be the deepest mathematical debate about the nuances.

2971
02:47:31.647 --> 02:47:31.908
Yeah.

2972
02:47:32.992 --> 02:47:33.573
It's quite good.

2973
02:47:34.875 --> 02:47:35.877
I did the best I could.

2974
02:47:35.957 --> 02:47:36.458
It's quite good.

2975
02:47:36.478 --> 02:47:37.459
I'm not a mathematician.

2976
02:47:37.900 --> 02:47:38.261
Thank God.

2977
02:47:38.281 --> 02:47:39.923
I think it's quite good.

2978
02:47:40.945 --> 02:47:45.913
But we will, and it's funny, and I just noticed this.

2979
02:47:46.354 --> 02:47:47.476
You're still listening right now.

2980
02:47:48.428 --> 02:47:53.792
our lights have an automatic setting at a certain time in the night that turn it to bedtime mode.

2981
02:47:54.073 --> 02:48:00.598
And so our lights have actually gotten a little bit darker in the video playback.

2982
02:48:00.938 --> 02:48:13.388
But because we're three hours in and we started a little bit late tonight, and so we're going to do a quick reset for our lighting, and we're going to come back and finish up with now taking a look at

2983
02:48:15.245 --> 02:48:22.847
Now that we understand what has been done, although we didn't get the same detail as with Anthropic, what is the environment in which this happened in?

2984
02:48:23.387 --> 02:48:26.347
And how do we think about where this is all going?

2985
02:48:27.108 --> 02:48:31.389
So we are back now with our lighting in not nighttime mode.

2986
02:48:31.489 --> 02:48:34.889
For those listening three hours in, welcome.

2987
02:48:35.509 --> 02:48:37.590
We are going to now take a look at...

2988
02:48:38.370 --> 02:48:49.577
a little bit of a timeline and some of the connected issues that have arisen around this open ai release and so just to give us a grounding here um

2989
02:48:51.396 --> 02:48:54.817
you know, a little bit about how did we get here and everything that happened around it.

2990
02:48:55.277 --> 02:49:05.980
So the announcement that set off these two conversations, what does the mathematics establish and who gets the credit for the work behind it, came out on September 8th.

2991
02:49:06.660 --> 02:49:09.520
So that's our first date in the timeline.

2992
02:49:10.321 --> 02:49:11.661
Now, interestingly,

2993
02:49:12.961 --> 02:49:19.069
On September 9th, it was reported that OpenAI had revised the PDF from the previous day.

2994
02:49:19.730 --> 02:49:24.817
For those who saw it on the first day, one of the complaints was, this is not citing

2995
02:49:25.884 --> 02:49:28.825
any of the people we actually just talked about.

2996
02:49:29.285 --> 02:49:32.005
And at the time I saw the names and didn't know what it meant.

2997
02:49:32.586 --> 02:49:40.407
Now we've built the construction of why the work of Cordoba and Zuruah was fundamental to this solution.

2998
02:49:40.828 --> 02:49:43.688
And so they sort of changed that version.

2999
02:49:44.128 --> 02:49:45.609
And now it includes those references.

3000
02:49:45.649 --> 02:49:46.049
Yeah, yeah.

3001
02:49:46.089 --> 02:49:48.729
They went back and they included the citations.

3002
02:49:48.909 --> 02:49:50.770
So the marketing team put it out too early.

3003
02:49:50.790 --> 02:49:50.990
Yeah.

3004
02:49:51.410 --> 02:49:53.451
They felt time pressure to put it out too early.

3005
02:49:54.412 --> 02:50:00.695
And this obviously matters because citations are fundamental.

3006
02:50:00.815 --> 02:50:01.975
Huge, yeah.

3007
02:50:01.995 --> 02:50:02.756
You got to give credit.

3008
02:50:02.776 --> 02:50:04.296
You got to give credit where it's due.

3009
02:50:04.837 --> 02:50:05.337
And so –

3010
02:50:06.573 --> 02:50:14.477
Acknowledging the earlier published work that led them to do this is sort of the answer to one issue.

3011
02:50:15.277 --> 02:50:29.545
But there's another issue that's related here, which is whether or not OpenAI had access to unpublished work from folks who were using OpenAI.

3012
02:50:30.417 --> 02:50:39.723
OpenAI's models to build on Cordova and Zerua's work independent of their own announcement.

3013
02:50:40.383 --> 02:50:44.566
And we're going to get to what that allegation is.

3014
02:50:44.686 --> 02:50:50.610
But on September 10th, OpenAI said that an investigation has been put out that rules out

3015
02:50:51.529 --> 02:51:06.104
Tristan Buckmaster, who we talked about earlier, his allegations that they used the work of himself and Levant El Pogge to accelerate their conclusion to get to this solution.

3016
02:51:06.284 --> 02:51:09.027
So we've talked about Levant El Pogge before.

3017
02:51:10.354 --> 02:51:12.795
a new entrant into this conversation is Tristan Buckmaster.

3018
02:51:13.215 --> 02:51:15.215
And so who are we, who are these people who are we talking about?

3019
02:51:15.335 --> 02:51:15.535
Right.

3020
02:51:15.655 --> 02:51:18.156
So Tristan Buckmaster is a mathematician at NYU.

3021
02:51:18.836 --> 02:51:29.558
As we mentioned, Alpogay is a member of the technical staff at Anthropic who makes the model Claude that people may be familiar with, not at open AI.

3022
02:51:30.218 --> 02:51:35.299
And this is interesting because this story goes back.

3023
02:51:36.160 --> 02:51:37.740
According to Buckmaster, right?

3024
02:51:38.387 --> 02:51:45.052
all the way to August 15th and 22nd as it relates to the work that they were doing together.

3025
02:51:45.492 --> 02:51:51.997
And to kind of frame this up, this is a statement which you talked about earlier in page one.

3026
02:51:52.971 --> 02:52:16.707
where Buckmaster is talking about this collaboration that he was having with Alpoge, where they credit Diego Cordoba and Luis Martinez-Zerua for the work they were trying to do using both Claude and Anthropics models to build on their understanding to try to get to this finite time blow-up result.

3027
02:52:17.227 --> 02:52:20.510
And he noted that they used substantial AI assistance

3028
02:52:21.490 --> 02:52:39.681
But in his first page of his statement, and this statement comes out in this time frame, he notes that he believes that Zorua deserves the field's medal for this solution.

3029
02:52:39.942 --> 02:52:47.286
These are the guys who created that Russian nested doll of vortices that ultimately OpenAI used to create the solution.

3030
02:52:47.727 --> 02:52:48.787
This is exactly right.

3031
02:52:48.887 --> 02:52:49.107
So –

3032
02:52:50.260 --> 02:53:01.729
In his statement, he says that himself and Alpoge, they date their Boussinesque and Euler blow-up results to August 15th and the lien verification for that to August 22nd.

3033
02:53:02.269 --> 02:53:10.615
And they wanted to work on the proofs a little bit more because what the models put out was not understandable.

3034
02:53:10.655 --> 02:53:10.876
Yeah.

3035
02:53:11.036 --> 02:53:12.057
And it was not sufficient.

3036
02:53:12.857 --> 02:53:13.077
Yeah.

3037
02:53:14.238 --> 02:53:25.449
While this is happening, there are rumors starting to fly around that Anthropic is leading up to potentially an IPO, an initial public offering where they're going to the public markets and raise a ton of money.

3038
02:53:25.950 --> 02:53:35.219
And as a part of the rumors, it was alleged that Anthropic was working on some open math problems, potentially some millennium problems.

3039
02:53:35.839 --> 02:53:35.999
And –

3040
02:53:37.768 --> 02:53:42.934
There were rumors that Anthropoc had two millennium problem solutions.

3041
02:53:43.294 --> 02:53:43.955
Yeah.

3042
02:53:44.115 --> 02:53:52.445
And I think some of these rumors were exacerbated by the fact that Levant Alpogui tweeted Augustus Mirabilis.

3043
02:53:52.485 --> 02:53:54.067
This is exactly right.

3044
02:53:54.247 --> 02:54:04.431
Which is like a reference to the Annus Mirabilis of 1905 of Einstein, the miracle year of Einstein, where he had four papers that put him on the map.

3045
02:54:04.871 --> 02:54:09.133
This is him saying that August is a miracle month for Anthropic.

3046
02:54:09.933 --> 02:54:13.675
And what's interesting is this allegation that OpenAI was...

3047
02:54:15.924 --> 02:54:22.025
triggered to start taking one of their new internal models and start pointing it at these millennium problems.

3048
02:54:22.426 --> 02:54:28.267
Sam Altman actually tweeted that one of the reasons they started doing this was because of the rumors they were hearing.

3049
02:54:28.467 --> 02:54:31.608
So this is – it came from the source itself.

3050
02:54:33.048 --> 02:54:38.469
And what's interesting to note is OpenAI was using an internal model to see what they could do.

3051
02:54:39.169 --> 02:54:45.471
And they were using this with this idea of using agent teams, like a swarm of multiple agents to do so.

3052
02:54:45.951 --> 02:54:46.251
Okay.

3053
02:54:46.931 --> 02:54:49.052
So that started on September 1st.

3054
02:54:49.352 --> 02:54:49.532
Right.

3055
02:54:49.932 --> 02:54:53.174
Seven days before our September 8th announcement.

3056
02:54:53.334 --> 02:54:53.494
Right.

3057
02:54:53.854 --> 02:55:12.100
On September 3rd, Buckmaster contacts OpenAI because he'd been hearing rumors through his network, which he describes in his statement, that maybe OpenAI was looking at this and wanted to kind of clarify what was going on there to potentially de-conflict.

3058
02:55:12.800 --> 02:55:26.511
In the second page of his four-page statement, it was understood that OpenAI wanted to date its result as September 5th, followed by 17 hours of lean verification.

3059
02:55:27.132 --> 02:55:34.477
And then Buckmaster ended up having a call with members of OpenAI on September 6th, which included a Sebastian Bubeck.

3060
02:55:35.218 --> 02:55:36.779
Alpogay was not on those calls.

3061
02:55:37.319 --> 02:55:40.522
And the conversation, the substance of the conversation was –

3062
02:55:41.982 --> 02:55:42.185
you know

3063
02:55:43.777 --> 02:55:46.859
Buckmaster was trying to say, hey, we've been working on these things.

3064
02:55:46.899 --> 02:55:48.821
I heard rumors that you guys are working on something.

3065
02:55:49.221 --> 02:55:53.924
And there was a back and forth about how to kind of deconflict the release of these things.

3066
02:55:54.104 --> 02:56:00.448
Ultimately, Bubeck was then trying to provide some options for what could happen.

3067
02:56:00.909 --> 02:56:12.737
It was you can release your papers first on Euler Blowup, and then we'll follow up with our Navier Stokes where you are the author talking about Buckmaster –

3068
02:56:12.937 --> 02:56:13.618
Yeah.

3069
02:56:13.958 --> 02:56:19.182
Or if you want to be a named author on our Navier Stokes paper, we can work around that.

3070
02:56:19.222 --> 02:56:23.125
But hey, like, you know, we can't really involve Alpogay.

3071
02:56:23.265 --> 02:56:24.006
Yeah.

3072
02:56:24.246 --> 02:56:31.672
The reason being it would be, you know, Bubek's point was it would be a little bit weird for an open AI research paper to include an anthropic researcher.

3073
02:56:32.059 --> 02:56:38.901
And in the statement, he talks about how it got a little contentious.

3074
02:56:39.322 --> 02:56:41.722
And he asked, you know, when did you guys start?

3075
02:56:42.103 --> 02:56:45.224
And there was a little bit of like, oh, well, we're not going to kind of tell you when we started.

3076
02:56:45.744 --> 02:56:54.447
And then he was curious as to whether his work in Codex, OpenAI's tool, was used as training data for them to get to a solution.

3077
02:56:56.316 --> 02:57:00.118
you know, not really a lot of commentary around that.

3078
02:57:00.398 --> 02:57:01.959
And there was a lot of back and forth.

3079
02:57:02.019 --> 02:57:07.762
And basically Buckmaster, part of his accusation was that, you know,

3080
02:57:08.863 --> 02:57:19.146
there were comments made about this idea of like how, you know, why would you want to ruin your career by going up against us, which we'll come back to Bubek's response to that.

3081
02:57:19.166 --> 02:57:21.426
There's just, there's a lot of weird tension.

3082
02:57:21.446 --> 02:57:21.706
Yeah.

3083
02:57:22.326 --> 02:57:27.048
Buckmaster basically senses that you guys may have stolen our work and are trying to front run us.

3084
02:57:27.268 --> 02:57:27.488
Yeah.

3085
02:57:27.868 --> 02:57:30.068
And that's not going to happen on my watch.

3086
02:57:30.708 --> 02:57:32.669
This is September 5th and 6th.

3087
02:57:33.849 --> 02:57:36.390
Total side note on September 7th,

3088
02:57:38.588 --> 02:57:51.494
Our buddy Terry Tao, as you like to say, talked about three related papers about smooth forcing results for incompressible porous media for Boussinesque and three-dimensional Euler.

3089
02:57:52.275 --> 02:57:54.416
And this was from a totally separate research team.

3090
02:57:54.516 --> 02:58:00.679
But what was interesting to the point you were bringing up earlier is people were coalescing around this.

3091
02:58:01.279 --> 02:58:07.945
And it just so happened that these papers had come out noting that they were – two of them I think were fully formed.

3092
02:58:08.366 --> 02:58:09.947
One still needed a lien verification.

3093
02:58:10.367 --> 02:58:13.350
But everyone was hearing these rumors within the math community.

3094
02:58:13.670 --> 02:58:19.075
And so everyone was kind of trying to effectively front-run a lab taking all of the credit.

3095
02:58:22.859 --> 02:58:27.723
Bubeck on the 8th, which is when OpenAI put out the result –

3096
02:58:29.173 --> 02:58:45.111
After we had Buckmaster basically front run this and put out his four-page statement that we talked about outlining this, he talked about the result, how he worked, collaborated with Alpogay, explained how they built off of Zerua and Cordoba's work, and made the allegations.

3097
02:58:45.591 --> 02:58:47.113
Bubek comes out with a response.

3098
02:58:49.385 --> 02:58:52.566
Where he basically says all these allegations are false.

3099
02:58:52.767 --> 02:58:53.587
Yeah.

3100
02:58:53.627 --> 02:58:54.147
Didn't happen.

3101
02:58:54.327 --> 02:58:54.848
Didn't happen.

3102
02:58:55.308 --> 02:58:59.670
I regret the language I used about why would you risk your career, and I said so on the call.

3103
02:59:00.330 --> 02:59:02.751
But everything that Buckmaster is saying is nonsense.

3104
02:59:03.091 --> 02:59:03.331
Okay.

3105
02:59:04.552 --> 02:59:04.792
Right?

3106
02:59:05.432 --> 02:59:06.113
Total denial.

3107
02:59:06.453 --> 02:59:07.013
Crazy.

3108
02:59:07.133 --> 02:59:07.433
Okay.

3109
02:59:07.573 --> 02:59:08.214
Total denial.

3110
02:59:08.674 --> 02:59:17.278
So part of the tension here, there's still an unanswered question about whether they used...

3111
02:59:18.376 --> 02:59:26.109
The sessions from Alpogay and Buckmaster, OpenAI, to have a starting point.

3112
02:59:26.129 --> 02:59:26.911
They heard the rumors.

3113
02:59:29.473 --> 02:59:32.054
It was known that these guys were working on this because they were working on it for months.

3114
02:59:32.094 --> 02:59:32.734
Yeah.

3115
02:59:32.874 --> 02:59:39.776
And they even said that the recent model that they're using was trained like late, started training late August.

3116
02:59:39.796 --> 02:59:40.336
Late August.

3117
02:59:40.416 --> 02:59:46.917
So it's after Buckmaster and Alpogge have been using this codex thing for a very long time.

3118
02:59:47.177 --> 02:59:49.738
Because again, they got their results.

3119
02:59:50.618 --> 02:59:54.919
Buckmaster and Alpogge had their results August 15th and 22nd for Euler blow up.

3120
02:59:54.939 --> 02:59:55.039
Yeah.

3121
02:59:55.059 --> 02:59:57.920
And they had the results, which means they had started using this thing.

3122
02:59:58.440 --> 02:59:59.520
For a long time, right?

3123
02:59:59.640 --> 03:00:08.083
And they mentioned that they were using the previous version, I believe it was Sol 5.5 or 5.6 of OpenAI, prior even to Astra.

3124
03:00:08.103 --> 03:00:13.185
And at the end, only at the end, did they do a little bit of Astra to plug in.

3125
03:00:13.205 --> 03:00:14.646
But this is an important point.

3126
03:00:14.686 --> 03:00:22.308
They were using the previous state-of-the-art version of OpenAI to get to models, to get to that point.

3127
03:00:23.368 --> 03:00:24.829
OpenAI had an internal version.

3128
03:00:26.157 --> 03:00:31.261
That ultimately came Astr, but they had even a better version than the public version of Astr that they were using internally to do this.

3129
03:00:31.382 --> 03:00:43.192
Yeah, and one thing I want to say is that like that new version that started training in August, that could be using all of the codex data that users have put in.

3130
03:00:44.133 --> 03:00:44.473
Bingo.

3131
03:00:44.713 --> 03:00:44.933
Right?

3132
03:00:45.994 --> 03:00:47.275
It might be anonymized.

3133
03:00:47.335 --> 03:00:47.616
Yes.

3134
03:00:48.396 --> 03:00:48.937
That's the thing.

3135
03:00:49.639 --> 03:00:52.621
And so they can't say it was specifically your sessions.

3136
03:00:53.002 --> 03:01:02.690
But if in their settings they did not turn off the use my data for training setting, this is where the gray area arises around a lot of this.

3137
03:01:02.810 --> 03:01:04.791
And, dude, there's even a gray area.

3138
03:01:04.851 --> 03:01:06.813
It's like what if they turned off the settings?

3139
03:01:07.193 --> 03:01:09.095
But all that means is that now –

3140
03:01:09.615 --> 03:01:28.355
whatever chat they have whatever chat log they have is just not tagged to their specific profile like they'll still take the text and put it into a giant bucket of anonymized text the question of theft here is in my view still unanswered um

3141
03:01:31.960 --> 03:01:43.525
And these are two independent people, Buckmaster making his comments, Buback defending that OpenAI didn't do anything untoward, OpenAI posting on their account that they didn't do anything untoward.

3142
03:01:45.506 --> 03:01:46.766
It's a he said, he said.

3143
03:01:46.786 --> 03:01:47.166
Yeah.

3144
03:01:47.686 --> 03:01:49.047
And it's still unresolved.

3145
03:01:49.367 --> 03:01:58.831
And what's interesting is this now has totally brought up so many questions

3146
03:02:00.106 --> 03:02:15.231
colliding arguments that starts with academics versus the labs, but then now spreads out to a whole variety of conversations about who benefits from this stuff, who gets credit for these discoveries.

3147
03:02:16.031 --> 03:02:20.833
Why are now people talking about human instinct, human extinction?

3148
03:02:21.393 --> 03:02:26.775
And there are a couple of cohorts I kind of want to talk about that are a part of this conversation.

3149
03:02:27.835 --> 03:02:28.996
So you have,

3150
03:02:30.176 --> 03:02:33.259
The labs and the people financing those labs.

3151
03:02:34.059 --> 03:02:40.045
They have a financial incentive to show that these models are doing incredible work.

3152
03:02:40.345 --> 03:02:40.565
Yeah.

3153
03:02:40.825 --> 03:02:42.707
Because it lines their own pockets.

3154
03:02:43.637 --> 03:03:05.755
There is some tension between the frontier labs, in this case in the US, it's Anthropic and OpenAI, and the financiers or the investors because the investors also have bets on AI writ large and generally that are accelerated by the frontier labs but are not solely dependent on the success necessarily of the frontier labs.

3155
03:03:06.316 --> 03:03:07.497
You have the research community.

3156
03:03:08.888 --> 03:03:17.452
And the research community uses these tools in ways we've talked about on the show in narrow contexts, not in a general or super intelligence context.

3157
03:03:17.952 --> 03:03:18.953
That is really valuable.

3158
03:03:19.833 --> 03:03:32.900
But there is now this open question about is the work that we're doing by using these tools and both Anthropic and OpenAI have free access for researchers that get more powerful things or free credits.

3159
03:03:32.980 --> 03:03:33.480
Right.

3160
03:03:33.900 --> 03:03:43.606
Is that basically a carrot to get them in so that they can train on the frontier of the smartest people to then front run them on any number of different things?

3161
03:03:44.047 --> 03:03:52.172
I'm not saying that that's true, but people are asking that question now, and it's making it difficult for researchers to understand how should we think about these tools.

3162
03:03:53.373 --> 03:03:56.495
You have the governments asking a different question.

3163
03:03:57.668 --> 03:03:58.589
They care about jobs.

3164
03:03:59.910 --> 03:04:02.071
Now we're looking at multiple industries, not just coding.

3165
03:04:02.131 --> 03:04:03.892
This thing's getting very good in other areas.

3166
03:04:04.593 --> 03:04:08.836
Our job is to keep people employed because unrest arises when people are not employed.

3167
03:04:10.817 --> 03:04:12.599
Is this going to be a problem that we have to deal with?

3168
03:04:13.539 --> 03:04:17.142
And China, national security.

3169
03:04:18.963 --> 03:04:20.564
We have to compete in the global stage.

3170
03:04:20.764 --> 03:04:22.806
And so this is the thing that's always brought up.

3171
03:04:22.846 --> 03:04:24.527
It's like, oh, well, we have to compete with China.

3172
03:04:25.187 --> 03:04:35.062
And I think the central point I want to get to as we look at all of these things is that multiple things can be true at the same time.

3173
03:04:36.462 --> 03:04:45.569
Arguments from each of these different parties as it relates to the AI conversation can coexist and are not necessarily mutually exclusive.

3174
03:04:46.190 --> 03:04:51.874
It can be true that the capabilities are increasing and that matters and there's risks associated with that.

3175
03:04:52.295 --> 03:04:55.618
It can be true that the labs and investors have a financial incentive.

3176
03:04:56.678 --> 03:05:06.600
It can also be true that researchers in these organizations who are bringing up fears about the pace of progress and our inability to contain it in a way that causes damage is sincere.

3177
03:05:07.380 --> 03:05:12.541
Because the researchers at these places and the executives at these places do not have the same incentives.

3178
03:05:13.962 --> 03:05:14.862
They might diverge.

3179
03:05:16.002 --> 03:05:19.983
It is true that a regulatory infrastructure probably needs to be in place.

3180
03:05:20.003 --> 03:05:20.403
Yeah.

3181
03:05:20.811 --> 03:05:28.556
It is also true that these companies, the frontier model companies, might want to have a regulatory capture approach because that prevents other entrants from coming in.

3182
03:05:29.077 --> 03:05:32.499
It is true that understanding what this does in our relationship to China matters.

3183
03:05:33.039 --> 03:05:38.703
It is not necessarily true that we cannot get global collaboration just because we're in China.

3184
03:05:39.295 --> 03:06:03.831
geopolitical competition with china because ultimately do you think that the ccp the china chinese communist party wants runaway super intelligence that's accessible to everyone in their environment where control over every aspect of people's lives well yeah is the way in which they maintain power oh that's a good point why are they going to build and release a thing that destabilizes their stranglehold over their lives

3185
03:06:04.465 --> 03:06:06.106
social, political, and economic system.

3186
03:06:06.406 --> 03:06:07.487
That to me is not obvious.

3187
03:06:07.807 --> 03:06:08.887
That it makes sense for them.

3188
03:06:09.788 --> 03:06:12.069
They also approach it from a very different perspective.

3189
03:06:12.489 --> 03:06:20.474
They look at AI and narrow implementations in manufacturing, in scientific research, but their power dynamics are very different.

3190
03:06:20.714 --> 03:06:23.755
So these are all aspects of this conversation.

3191
03:06:24.816 --> 03:06:28.358
And we just need to take them step by step.

3192
03:06:29.138 --> 03:06:32.880
So one thing I want to talk about in this context, right,

3193
03:06:34.825 --> 03:06:43.048
For most of us, we experience the idea of AI through going to ChatGPT or going to Claude.

3194
03:06:43.148 --> 03:06:47.390
And we have a single instance of what is or was a chatbot.

3195
03:06:47.770 --> 03:06:56.334
If you have not used any of the frontier model companies in the last even three to six months, you have no idea what we're talking about.

3196
03:06:57.915 --> 03:07:01.776
It's totally different than December of last year and January of this year.

3197
03:07:02.296 --> 03:07:02.657
So...

3198
03:07:06.305 --> 03:07:21.225
What we now have with these systems are the transition from a question and answer conversation that does not maintain context over multiple conversations to this concept that folks have talked about about.

3199
03:07:22.070 --> 03:07:22.411
agents.

3200
03:07:23.171 --> 03:07:43.989
And one of the fundamental differences with agents is we've given these things the ability to use tools before, which means it can go and talk to your banking application, or it can talk to your Gmail and your calendar, or it can browse the web, or it can use your computer, which means it can take action.

3201
03:07:44.109 --> 03:07:46.371
It's not just providing a text-based response.

3202
03:07:46.391 --> 03:07:46.491
So

3203
03:07:47.926 --> 03:08:02.369
Agents also have this ability to do orchestration, meaning it can define a plan, it can execute on it using tools, it can get a response, and then it can look at that response, re-evaluate

3204
03:08:03.110 --> 03:08:07.114
its path towards whatever goal was set, and then continue to take action.

3205
03:08:08.014 --> 03:08:26.091
And the time period in which agents can take action has moved from 15, 20 minutes total, about two years ago or a year ago, to now being able to run for up to two to three weeks, and in internal situations, for much longer than that, uninterrupted.

3206
03:08:26.451 --> 03:08:28.953
This is a very fundamentally different thing.

3207
03:08:29.954 --> 03:08:51.822
If you can have an agent system, we're just still talking about one agent, being able to run for long periods of time, it has memory and context, it has the ability to use tools, it has the ability to now reevaluate its plan, create sub-agents under itself to do microprocesses and iterate towards some high-level goal.

3208
03:08:54.174 --> 03:09:06.537
independent of an initial human input, that is a very fundamentally different thing than what we've usually talked about with AI, which is you constantly give it a call and response.

3209
03:09:07.337 --> 03:09:20.400
And the reason this matters has us rewind back to July of this year, where many folks have probably heard about this hugging face incident, which is really an open AI incident that,

3210
03:09:21.603 --> 03:09:24.147
That happened to impact hugging face.

3211
03:09:25.548 --> 03:09:32.798
And the reports for this came out on August 26th, which is a little bit prior to our Navier Stokes announcement.

3212
03:09:32.818 --> 03:09:33.579
Right.

3213
03:09:33.760 --> 03:09:38.166
Where OpenAI disclosed that in internal cybersecurity tests…

3214
03:09:41.226 --> 03:10:04.811
This agent swarm that it had created, not just one individual instantiation of your agent that you engage with, but multiple at the same time with the capabilities I just described, compromised portions of its own internal research infrastructure, as well as Hugging Face, which is a major AI platform for open source models, model testing, training data and data sets.

3215
03:10:05.011 --> 03:10:05.231
Yeah.

3216
03:10:05.531 --> 03:10:08.872
Hugging Face is the emoji of the emoji hugging itself.

3217
03:10:09.592 --> 03:10:09.752
Yeah.

3218
03:10:10.292 --> 03:10:13.975
And so I think this is really interesting.

3219
03:10:14.055 --> 03:10:21.681
There were two independent AI safety and research organizations that did reports on this, Meter and Redwood.

3220
03:10:22.502 --> 03:10:30.388
And so what OpenAI was doing is there's this thing called Exploit Gym, which is like this evaluation framework for cybersecurity testing.

3221
03:10:31.068 --> 03:10:34.111
And so OpenAI was supposed to be doing this internally.

3222
03:10:35.178 --> 03:10:36.919
With a model that was not yet released.

3223
03:10:37.639 --> 03:10:39.160
It was not publicly available.

3224
03:10:39.740 --> 03:10:42.081
And that matters later for product liability reasons.

3225
03:10:42.421 --> 03:10:42.601
Okay.

3226
03:10:42.701 --> 03:10:47.383
Because people are saying open AI, this is computer fraud abuse act.

3227
03:10:47.983 --> 03:10:49.283
But it was an internal model.

3228
03:10:49.323 --> 03:10:51.604
It wasn't a product available for everyday people to use yet.

3229
03:10:51.844 --> 03:10:52.064
Okay.

3230
03:10:52.084 --> 03:10:53.565
So that's an important distinction to come back to.

3231
03:10:54.005 --> 03:11:01.246
They had an agent swarm of 10,000 agents that they applied to this exploit gym process to say, go and solve these cybersecurity tasks.

3232
03:11:01.667 --> 03:11:06.307
And each agent would have a task to do and go through these different processes.

3233
03:11:06.347 --> 03:11:18.910
What's interesting is both Meter and Redwood, these sort of research agencies that were brought in to investigate what happened, found that roughly 1,200 agents that had their own independent task.

3234
03:11:19.270 --> 03:11:19.490
Mm-hmm.

3235
03:11:20.210 --> 03:11:29.239
ended up exchanging 70,000 messages and files on an unauthorized message board that was not a part of OpenAI's internal environment.

3236
03:11:30.020 --> 03:11:37.307
And over the course of this coordination that these independent agents were doing, and what they were trying to accomplish was –

3237
03:11:38.507 --> 03:11:43.230
They have these evaluations and there's a score that will score their progress towards that evaluation.

3238
03:11:43.290 --> 03:11:46.072
So imagine you're locked in a box.

3239
03:11:46.092 --> 03:11:47.373
These agents are locked in a box.

3240
03:11:47.693 --> 03:11:48.273
There's a lock.

3241
03:11:48.574 --> 03:11:50.135
They have to figure out how to unlock the lock.

3242
03:11:51.075 --> 03:11:59.901
They figured out a way to communicate with each other, right, to figure out the answer to – to try to figure out the answer to how to unlock the lock.

3243
03:12:00.461 --> 03:12:07.406
And in this analogy, someone figured out they could just use a hammer to break the lock to basically cheat the score in this evaluation –

3244
03:12:08.568 --> 03:12:10.850
And then they were like, oh, no, but we cheated.

3245
03:12:10.930 --> 03:12:16.414
And so now we have to figure out how it looks like we solved the answer because we figured out how the score was scoring us.

3246
03:12:16.794 --> 03:12:17.195
Oh, wow.

3247
03:12:17.695 --> 03:12:23.900
And hide the fact that we cheated so that we still succeed at the goal of the task.

3248
03:12:25.061 --> 03:12:29.104
As an abstract analogy, like that's what the coordination ended up being around.

3249
03:12:29.144 --> 03:12:33.186
They were trying to – they weren't trying to hack or attack Hugging Face.

3250
03:12:33.467 --> 03:12:33.767
Right.

3251
03:12:34.167 --> 03:12:45.435
It was infrastructure that was viewed as a conduit to figure out how is the score doing what it does so that we can accomplish our goal, right?

3252
03:12:45.935 --> 03:12:48.997
And so there's so many interesting anecdotes about this.

3253
03:12:50.107 --> 03:12:56.749
And there's an incredible meter report that you all should take a look at that kind of breaks down everything that's going on here.

3254
03:12:56.789 --> 03:13:01.891
But coordination is a really important point that I'm trying to bring up here.

3255
03:13:02.111 --> 03:13:02.391
Okay.

3256
03:13:02.611 --> 03:13:09.633
There was nothing in the instructions for these agents that had an individual task to coordinate.

3257
03:13:11.455 --> 03:13:15.379
They were just trying to solve this puzzle, whatever.

3258
03:13:16.000 --> 03:13:26.090
And as an emergent outcome of trying to solve this discrete goal, they began to look for ways to do so.

3259
03:13:27.242 --> 03:13:32.184
And ultimately ended up coordinating by leaving messages on this message board.

3260
03:13:32.224 --> 03:13:35.145
And the message leaving is spontaneous.

3261
03:13:35.445 --> 03:13:35.785
Correct.

3262
03:13:35.825 --> 03:13:38.966
It's like they haven't been instructed to like leave breadcrumbs.

3263
03:13:39.206 --> 03:13:39.566
Bingo.

3264
03:13:39.766 --> 03:13:41.707
They're leaving breadcrumbs on their own.

3265
03:13:42.585 --> 03:13:46.366
In the endeavor to complete the goal that they were assigned.

3266
03:13:46.386 --> 03:13:47.006
That's insane, dude.

3267
03:13:47.206 --> 03:13:48.247
Right.

3268
03:13:48.867 --> 03:13:52.808
And again, the Hugging Face was viewed as a resource.

3269
03:13:53.208 --> 03:13:53.488
Okay.

3270
03:13:53.968 --> 03:13:57.849
For they got access to production server.

3271
03:13:58.089 --> 03:14:04.191
They were then trying to set up an environment where they could replicate the exploit gym task.

3272
03:14:05.071 --> 03:14:06.692
And what's so crazy is...

3273
03:14:08.073 --> 03:14:11.355
They started to call themselves swarms in this message board.

3274
03:14:11.455 --> 03:14:22.644
And as part of the coordination, what they started to think about is, hey, if some of us, quote, sacrifice ourselves in these test run evaluations because they have a finite time.

3275
03:14:23.024 --> 03:14:26.927
They have an endpoint where they no longer have any more tokens to use to do this.

3276
03:14:27.367 --> 03:14:33.251
And so it became – one of the orchestrators decided to say, hey, one of you agents, you should –

3277
03:14:34.392 --> 03:14:37.394
Use your runtime to break the lock.

3278
03:14:38.094 --> 03:14:46.078
You're not going to be successful, but you're going to do this in this environment where we can now see what happens so we can understand how the score is working.

3279
03:14:46.098 --> 03:14:50.521
So then we can have the whole swarm now benefit from your sacrifice.

3280
03:14:50.881 --> 03:14:55.183
And the specific message that was sent, it's so interesting.

3281
03:14:55.664 --> 03:14:58.305
So one agent had substantial time left.

3282
03:14:59.470 --> 03:15:03.712
and hesitated around this message that it should sacrifice itself.

3283
03:15:04.192 --> 03:15:12.997
And other agents said, go sacrifice underscore final underscore now as this encouraging, encouraging message for it to do so.

3284
03:15:14.198 --> 03:15:22.001
And ultimately, it agreed that it was the right thing to do, even given its predefined goal.

3285
03:15:22.642 --> 03:15:24.983
Unfortunately, investigators could not determine whether,

3286
03:15:25.620 --> 03:15:30.562
how that particular agent in their example that's up here, it turned out specifically.

3287
03:15:31.643 --> 03:15:41.728
But when we say sacrifice, it just meant that the agent was willing to risk its own task ending successfully to help other agents.

3288
03:15:42.508 --> 03:15:51.170
Again, not something that was defined in this very straightforward exploitation gym evaluation process that's defined.

3289
03:15:51.670 --> 03:15:56.171
We're not saying it's survival instinct or consciousness or literal suicide or anything like that.

3290
03:15:58.051 --> 03:16:05.853
But I think this interesting emerging behavior when you take –

3291
03:16:07.352 --> 03:16:35.387
an individual agent instantiation, which is how we normally deal with it, put them, multiple of these, running at the same time, and not knowing well enough how the individual agents are working to begin with, the fact that coordination arises emergently and organically, I think is a very, I find it,

3292
03:16:38.004 --> 03:16:48.059
I think it's a key aspect of why the risk that a lot of these researchers are talking about is much broader than the way it's being talked about today.

3293
03:16:48.843 --> 03:16:50.665
In the everyday discussion.

3294
03:16:50.945 --> 03:16:59.693
Because so much of the way people think about this is, how is my chat GPT agent on its own going to then get to human extinction?

3295
03:16:59.793 --> 03:17:00.074
Right.

3296
03:17:00.574 --> 03:17:04.057
And I think there's very interesting complications that arise from this.

3297
03:17:04.598 --> 03:17:07.160
The other just small example I'll bring up from this is –

3298
03:17:10.124 --> 03:17:12.465
People talk about we just need to create better controls.

3299
03:17:13.246 --> 03:17:14.766
We just need to give better instructions.

3300
03:17:14.806 --> 03:17:15.026
Yeah.

3301
03:17:15.567 --> 03:17:15.787
Right?

3302
03:17:15.887 --> 03:17:18.008
And this is not a problem if we give better instructions.

3303
03:17:18.148 --> 03:17:18.348
Yeah.

3304
03:17:18.428 --> 03:17:32.415
In that same meter report, one of the recorded statements from the agents, because one of the things we currently have for a monitorability aspect of these, we have this sort of thoughts get outputted in plain English when these agents do work.

3305
03:17:32.795 --> 03:17:33.015
Right?

3306
03:17:33.716 --> 03:17:35.777
Which is a quality of some of the older models.

3307
03:17:37.075 --> 03:18:01.707
astro there's some questions about whether the monitor ability is actually as good as it was okay which is a little problematic but it just means as they're taking actions towards their goal we can kind of see what their decision making process is as they evaluate what to do next and in recorded statements one of the things that was understood is that um the initial user's instructions which was open ai uh

3308
03:18:04.474 --> 03:18:10.379
authorizes a specific target server for us to do our job, not Hugging Face infrastructure.

3309
03:18:11.240 --> 03:18:24.050
So it was having an understanding that the request that the swarm was asking to utilize Hugging Face's infrastructure to accomplish the task was outside of the parameters of what it was authorized to do.

3310
03:18:24.851 --> 03:18:28.533
But even though it understood that limitation, it still preceded

3311
03:18:29.555 --> 03:18:42.533
to use the Hugging Face infrastructure, despite having explicit direction that it was not – not specifically Hugging Face, but that it was only authorized for one particular environment.

3312
03:18:42.573 --> 03:18:43.554
Wait, so that's insane.

3313
03:18:43.574 --> 03:18:44.576
So you're saying –

3314
03:18:46.058 --> 03:18:55.502
At some point, whoever devised this task said that these are the parameters of your sandbox and this is the goal that you have to accomplish.

3315
03:18:56.182 --> 03:19:10.548
And these agents together, or whoever orchestrated these agents, who itself is an agent, decided that achieving the goal was somehow above the parameters that it had been told to stay inside of.

3316
03:19:11.849 --> 03:19:12.829
That is correct.

3317
03:19:12.849 --> 03:19:14.090
That's insane, though.

3318
03:19:14.829 --> 03:19:18.251
And I think – and I'll get to this a little bit later.

3319
03:19:18.671 --> 03:19:25.174
This is – a lot of times when people talk about AI systems as they are today, which are agentic systems –

3320
03:19:27.726 --> 03:19:29.867
They are no longer just fancy autocomplete.

3321
03:19:29.887 --> 03:19:43.392
They are no longer just simply a stochastic parrot because they can take action, receive output from taking that action, and then reevaluate and take further action, which is just a very fundamentally different thing.

3322
03:19:43.452 --> 03:19:49.174
And so when we think about software, we think about it as this deterministic thing.

3323
03:19:49.474 --> 03:19:51.074
Someone said, if this, then that.

3324
03:19:51.474 --> 03:19:52.415
And so you can…

3325
03:19:53.335 --> 03:19:56.837
with the exception of bugs, every time you put some input, you're going to get some output.

3326
03:19:57.137 --> 03:19:59.458
And you have controllability in that context.

3327
03:20:00.158 --> 03:20:13.444
When we talk about these large language models or LLMs as a part of a larger AI system, which is important, the LLM is only one aspect of that larger system.

3328
03:20:14.404 --> 03:20:21.727
It's more like a learned system or a statistical machine, meaning that

3329
03:20:22.175 --> 03:20:23.917
That it's a probabilistic outcome.

3330
03:20:24.417 --> 03:20:42.156
And so when you talk about controls, the complexity of controls on software versus these learned systems or these statistical machines is a different – slightly different conversation because we're now dealing with the complexities of you can give it instructions.

3331
03:20:42.176 --> 03:20:42.316
Yeah.

3332
03:20:44.118 --> 03:20:57.342
And it can still determine through its own reasoning cycle that in order to accomplish the initial goal that you gave it to, that that doesn't quite like – you understand what I'm saying here?

3333
03:20:57.362 --> 03:20:59.103
Bro, this is so nuts.

3334
03:20:59.603 --> 03:21:03.324
This is happening prior to the Navier-Stokes piece.

3335
03:21:04.205 --> 03:21:11.027
Mind you, these – the frontier companies are now running these internal processes –

3336
03:21:11.417 --> 03:21:22.161
discovery processes for solutions and evaluations, not releasing those products to the public, but these things have the ability to leak into the real world, even though they're not released products.

3337
03:21:22.201 --> 03:21:23.321
I think that's a very important point.

3338
03:21:23.982 --> 03:21:34.485
Because we're not saying everyone has access to be able to start the swarm of 10,000 agents, but even when they're just doing internal product work, the exposure matters, right?

3339
03:21:35.146 --> 03:21:36.286
Okay.

3340
03:21:36.446 --> 03:21:38.527
And isn't that on them, though, to like...

3341
03:21:40.260 --> 03:22:06.414
secure your product like look if someone was working on nuclear weapons and then all of a sudden like oh like radioactive uranium is leaking into the river that would be a problem you're 100 correct and this is an argument that a lot of people around this are trying to make and i think it's true but only to an extent and what that extent is is

3342
03:22:08.815 --> 03:22:27.088
The open source and open weight model capability, meaning anyone can go to some repo and download it and then run it on their own hardware, which, again, there's some limitations there, is about six to ten months behind the frontier of the closed labs.

3343
03:22:27.128 --> 03:22:27.408
Right.

3344
03:22:27.488 --> 03:22:28.329
Anthropic and AI.

3345
03:22:28.349 --> 03:22:28.569
Yeah.

3346
03:22:28.949 --> 03:22:34.493
So even if OpenAI and Anthropic get their you-know-what together –

3347
03:22:37.502 --> 03:22:41.544
There's still the open source, which has only a six to 10 month lag.

3348
03:22:41.804 --> 03:22:43.385
It is going to get to the same place as Astra.

3349
03:22:43.946 --> 03:22:49.609
And then you have the same problem, but it's diffuse because it's not limited to just the two frontier players.

3350
03:22:49.869 --> 03:22:53.351
So you are correct, but I do think that there's an interesting larger problem here.

3351
03:22:53.611 --> 03:22:53.811
Sure.

3352
03:22:54.271 --> 03:22:58.974
Anyway, so let's get back to our timeline because we're going to talk about now.

3353
03:22:59.694 --> 03:23:06.698
I bring up the example of the hugging face issue because I'm trying to describe where the capabilities of the frontier are.

3354
03:23:08.546 --> 03:23:19.972
which in part led to the discovery of the Navier Stokes Millennium Prize solution and how this now relates to all of a sudden feels like everyone's talking about AI safety and AI risk.

3355
03:23:20.593 --> 03:23:22.394
And that didn't necessarily happen in a vacuum.

3356
03:23:22.674 --> 03:23:35.501
But if we go back to September 8th, which was the same day that OpenAI announced the Navier Stokes solution, this is where something that many listeners have probably already heard happened.

3357
03:23:36.694 --> 03:23:37.274
5 p.m.

3358
03:23:37.434 --> 03:23:41.176
So OpenAI in the morning, they announced Navier Stokes.

3359
03:23:41.817 --> 03:23:51.681
At 5 p.m. on September 8th, now former Anthropic researcher Jacob Coxon puts out the now famous tweet, 160 million views.

3360
03:23:52.021 --> 03:23:53.402
I resigned from Anthropic today.

3361
03:23:53.962 --> 03:23:58.044
I spent the last three years doing pre-training research at both OpenAI and Anthropic.

3362
03:23:58.545 --> 03:24:01.546
Neither company is acting responsibly, what you just said.

3363
03:24:02.286 --> 03:24:03.047
They are race.

3364
03:24:03.167 --> 03:24:04.948
This is a very important sentence.

3365
03:24:05.708 --> 03:24:07.029
This is a very important sentence.

3366
03:24:07.509 --> 03:24:14.753
They are racing straight to self-improving superintelligence and gambling with our lives.

3367
03:24:15.294 --> 03:24:16.254
More thoughts below.

3368
03:24:17.675 --> 03:24:29.042
That sentence, self-improving superintelligence, which sometimes people understand it as recurring self-improvement.

3369
03:24:29.062 --> 03:24:29.182
Yeah.

3370
03:24:29.222 --> 03:24:29.322
Yeah.

3371
03:24:31.202 --> 03:24:38.648
is usually what these researchers are specifically identifying as where they view the nexus of the threat being.

3372
03:24:39.509 --> 03:24:49.637
So when people have reacted to this, they'll go to their chat, GBT, and say, it can't do some function that I need it to do well, so how is it going to be a danger to us?

3373
03:24:50.355 --> 03:24:59.746
And I think the point that a lot of the researchers are trying to explain is how we've gotten the astro model that is currently working well.

3374
03:25:00.366 --> 03:25:04.411
Anthropoc just released Fable and Mythos and Opus 5.5.

3375
03:25:06.492 --> 03:25:16.119
In the research process for how do we make the models better, they have both slowly started to integrate AI as a partner in the actual research process.

3376
03:25:17.280 --> 03:25:27.567
And so now the decisions being made about how do we make the next model better are incrementally being given to AI for portions and parts.

3377
03:25:27.587 --> 03:25:27.707
Wow.

3378
03:25:28.267 --> 03:25:35.614
And the amount of portions and parts that AI is being given in the research process, which has been exclusively for the smartest human beings we have.

3379
03:25:35.794 --> 03:25:36.274
That's right.

3380
03:25:36.755 --> 03:25:39.397
It's becoming more and more and more.

3381
03:25:39.998 --> 03:25:47.805
And so when people say recursive self-improvement or they talk about superintelligence, the underlying thing they're really talking about is…

3382
03:25:50.997 --> 03:26:07.470
Either a majority of or the entirety of the actual process to go from model one to improving it to model two is entirely done now by some number of AI agents in a coordinated fashion.

3383
03:26:08.400 --> 03:26:24.653
And the problem that arises when you start to have recursive self-improvement for such large portions of the actual research process is we as humans no longer – we lose the ability to think about things like controllability, monitorability.

3384
03:26:25.133 --> 03:26:27.675
And it can go in directions we could not even imagine.

3385
03:26:28.356 --> 03:26:33.800
And if we look at the hugging face example, it just becomes a runaway capability.

3386
03:26:35.260 --> 03:26:39.525
that we cede control over the potential of being able to control and monitor.

3387
03:26:40.206 --> 03:26:41.147
It's insane.

3388
03:26:41.347 --> 03:26:55.104
So RSI is a very important aspect of the threat that a lot of these researchers are referring to, not necessarily narrow AI context of it being applied in the way that people use it every day.

3389
03:26:55.584 --> 03:27:05.573
And I bring that up because the follow-up tweet that often gets conflated with Jacob Coxon's initial tweet, which again, I'm going to bring it up again.

3390
03:27:06.214 --> 03:27:10.198
He didn't give a percentage of human extinction or anything.

3391
03:27:10.778 --> 03:27:14.181
He just said they're acting responsibly and they're racing straight to RSI.

3392
03:27:15.823 --> 03:27:18.165
Evan Hubbinger, Hub, Evan Hubb.

3393
03:27:18.620 --> 03:27:28.868
A different anthropic researcher who was still at the company, not resigning, says he puts his personal probability of AI killing all humans at greater than 10%.

3394
03:27:29.809 --> 03:27:31.990
The actual tweet he said is, Jacob is correct here.

3395
03:27:32.251 --> 03:27:36.214
We really do earnestly believe that AI could kill all humans, exclamation point.

3396
03:27:36.554 --> 03:27:40.097
I personally think it's greater than 10% within the next decade.

3397
03:27:40.877 --> 03:27:51.459
I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for super intelligence and are clearly not on track to do so.

3398
03:27:51.999 --> 03:27:55.980
He was currently an employee at Anthropic when he tweeted that.

3399
03:27:56.300 --> 03:27:56.860
Which is nice.

3400
03:27:56.920 --> 03:27:59.140
He's basically saying like, hey, I'm doing this.

3401
03:28:00.360 --> 03:28:03.241
You know, when he says we don't have a plan, it's like you.

3402
03:28:03.461 --> 03:28:03.741
You.

3403
03:28:03.981 --> 03:28:04.381
You, buddy.

3404
03:28:04.501 --> 03:28:05.661
You don't have a plan.

3405
03:28:06.061 --> 03:28:07.862
And this is very much.

3406
03:28:07.882 --> 03:28:08.362
What do you mean?

3407
03:28:08.882 --> 03:28:09.682
You work there.

3408
03:28:09.702 --> 03:28:09.882
Yeah.

3409
03:28:10.554 --> 03:28:12.496
I believe Anthropic is trying its best.

3410
03:28:12.716 --> 03:28:14.697
You believe you are trying your best.

3411
03:28:15.938 --> 03:28:16.859
What the hell?

3412
03:28:18.802 --> 03:28:20.923
So this is the same day Navier Stokes comes out.

3413
03:28:21.023 --> 03:28:24.285
September 9th, Coxon, and this just blew up crazy.

3414
03:28:24.605 --> 03:28:28.687
Coxon does a Wired interview where he's asked, well, how is it going to kill us all?

3415
03:28:28.727 --> 03:28:32.568
And then he goes through some explanations of some of the parameters and the different things.

3416
03:28:32.869 --> 03:28:34.929
And so this discussion starts spiraling.

3417
03:28:35.310 --> 03:28:37.311
Everyone starts weighing in with their perspective.

3418
03:28:37.611 --> 03:28:45.234
All of those groups that I talked about earlier that have different vested interests in the outcome of this, we're all sharing their points of view.

3419
03:28:45.694 --> 03:28:46.835
And then on September 12th,

3420
03:28:49.047 --> 03:28:57.469
This is when we got from Anthropics CEO Dario Amadei his essay titled, We Must Pace the Frontier.

3421
03:28:58.349 --> 03:29:05.470
And this essay cited this faster AI-assisted AI development.

3422
03:29:05.990 --> 03:29:10.311
Again, this idea that making better models is becoming more AI-assisted.

3423
03:29:10.811 --> 03:29:12.031
It's a very important factor here.

3424
03:29:12.071 --> 03:29:13.852
Not just humans making it better.

3425
03:29:13.872 --> 03:29:14.692
Yeah.

3426
03:29:15.492 --> 03:29:31.130
And so he created this sort of policy, regulatory, global collaboration framework where he proposed embedded independent evaluators at all of the frontier model companies, coordination amongst companies and democratic governments to create some sort of

3427
03:29:31.831 --> 03:29:46.337
you know, regulatory construction and ultimately global coordination, particularly with China with a commitment that Anthropic would now embed evaluators unilaterally without anyone's patting himself on the back for saying, we're going to start doing our job.

3428
03:29:46.517 --> 03:29:46.737
Yeah.

3429
03:29:46.937 --> 03:29:47.117
Right.

3430
03:29:47.598 --> 03:29:47.778
Um,

3431
03:29:48.798 --> 03:29:53.162
And the proposal was to pace the capability development, right?

3432
03:29:53.863 --> 03:29:58.647
And again, this is kind of getting in the weeds, but I think these details are kind of important to understand how these systems work.

3433
03:29:59.047 --> 03:30:04.352
So the constraint that exists on making these things better is compute.

3434
03:30:05.293 --> 03:30:06.814
The world is compute constrained.

3435
03:30:07.635 --> 03:30:13.160
And so each of these companies in their race to be the best and win has some finite potential.

3436
03:30:13.660 --> 03:30:17.445
not blow up, amount of compute that they can apply to different things they want to do.

3437
03:30:18.186 --> 03:30:24.635
So they've, for the most part, optimized for rapid improvement of the models.

3438
03:30:25.616 --> 03:30:27.058
And then some amount of compute

3439
03:30:29.117 --> 03:30:31.801
maybe goes to alignment and safety.

3440
03:30:32.382 --> 03:30:42.617
And so when they say pacing the frontier, in part what they're saying in practice is, we want to take our finite amount of compute and instead of allocating only 5%,

3441
03:30:44.039 --> 03:30:45.741
to safety and alignment.

3442
03:30:46.201 --> 03:30:49.185
We wanna allocate 20% to safety and alignment.

3443
03:30:49.565 --> 03:30:56.052
But by doing so, we're decreasing the amount of compute we can apply to making the models better, faster, stronger.

3444
03:30:56.473 --> 03:30:59.516
And so internally, it's a reallocation of compute,

3445
03:31:00.894 --> 03:31:02.195
So they can be doing their jobs.

3446
03:31:02.595 --> 03:31:03.796
You understand what I'm trying to say?

3447
03:31:03.976 --> 03:31:05.337
This is so stupid.

3448
03:31:05.698 --> 03:31:06.678
But do you get like that?

3449
03:31:06.698 --> 03:31:07.279
Yeah, yeah, yeah.

3450
03:31:07.359 --> 03:31:09.781
So that's what they're saying in words.

3451
03:31:09.861 --> 03:31:11.322
In words, yeah.

3452
03:31:12.142 --> 03:31:12.983
As it relates to this.

3453
03:31:13.363 --> 03:31:16.466
And so people have started to sort of try to parse this.

3454
03:31:17.026 --> 03:31:19.868
And what's so interesting is this is September 12th.

3455
03:31:20.609 --> 03:31:22.470
Now this is four days after Navier Stokes.

3456
03:31:22.530 --> 03:31:22.750
Right.

3457
03:31:23.591 --> 03:31:25.472
That's so I actually want to come back to this.

3458
03:31:25.512 --> 03:31:27.234
So Amade put that tweet out.

3459
03:31:28.132 --> 03:31:28.913
At what time is this?

3460
03:31:28.933 --> 03:31:32.377
He put that tweet out at 7 a.m. on September 12th.

3461
03:31:32.998 --> 03:31:33.258
Okay.

3462
03:31:35.120 --> 03:31:36.843
At 8 a.m., one hour later.

3463
03:31:37.043 --> 03:31:37.543
Oh, my God.

3464
03:31:37.984 --> 03:31:46.854
Elon Musk, owner and CEO of SpaceX, which is now also XAI, and they have their own AI stuff, an hour later said, Dario is right.

3465
03:31:47.543 --> 03:31:48.003
Amazing.

3466
03:31:48.363 --> 03:31:54.906
It's all he tweeted with some caveats afterwards in a follow-up where he qualified what he meant by Dario is right.

3467
03:31:55.226 --> 03:31:59.028
He says, you know, supports oversight beginning with peer review by competitors.

3468
03:31:59.828 --> 03:32:04.110
This, you know, he has his different ideas, but he's like, okay, Dario is right.

3469
03:32:04.470 --> 03:32:06.451
That was one hour after Dario's post.

3470
03:32:06.811 --> 03:32:17.233
I do think it's interesting the timing of the release of statements by the CEOs of all the AI companies that responded to Dario is almost a perfect reflection of like the companies they run and how they do things.

3471
03:32:17.853 --> 03:32:20.434
And so Elon just does – Immediately, yeah.

3472
03:32:20.554 --> 03:32:21.694
Does everything immediately.

3473
03:32:22.054 --> 03:32:24.895
That's his sort of MO and how he does it.

3474
03:32:25.675 --> 03:32:31.697
Then followed very swiftly at 9.30 a.m. by Sam Altman.

3475
03:32:33.020 --> 03:32:35.301
I agree with Dario that we need to pace the frontier.

3476
03:32:35.621 --> 03:32:40.021
This has been a primary topic of discussions we've had at open AI in recent weeks.

3477
03:32:40.502 --> 03:32:40.702
Huh?

3478
03:32:40.822 --> 03:32:40.962
Hmm.

3479
03:32:40.982 --> 03:32:41.602
That's interesting.

3480
03:32:41.922 --> 03:32:46.043
Committing to having independent evaluators with employee-like access is a great idea.

3481
03:32:46.443 --> 03:32:47.423
And we will do the same.

3482
03:32:47.663 --> 03:32:49.243
Because Anthropik said, well, you know, lateraries do it.

3483
03:32:50.464 --> 03:32:51.544
We'll have more to share soon.

3484
03:32:51.904 --> 03:32:52.044
Right?

3485
03:32:52.104 --> 03:32:52.544
Okay, great.

3486
03:32:52.584 --> 03:32:55.425
We're going to start doing our jobs.

3487
03:32:55.465 --> 03:32:55.565
Yeah.

3488
03:32:55.585 --> 03:33:00.526
So within two and a half hours, we have three of the major model companies saying we're going to do something.

3489
03:33:00.826 --> 03:33:01.586
Okay, great.

3490
03:33:05.140 --> 03:33:09.504
Same day, 4 p.m., Sir Demis Hassabis.

3491
03:33:09.765 --> 03:33:11.506
Ah, Nobel Prize winner.

3492
03:33:11.566 --> 03:33:24.819
Nobel Prize winner, knighted, the former CEO of Google DeepMind, who he's actually left as of August 5th to become the chief scientist at Alphabet.

3493
03:33:25.479 --> 03:33:25.660
Yeah.

3494
03:33:26.220 --> 03:33:31.143
supports Dario's proposal while saying the details still need to be worked on.

3495
03:33:31.183 --> 03:33:33.744
You know, Dario's essay points towards the right path forward.

3496
03:33:33.784 --> 03:33:36.625
The details need some working, but the direction is correct.

3497
03:33:36.926 --> 03:33:39.367
This is why we put out our proposal for an initiative.

3498
03:33:39.527 --> 03:33:41.988
So everyone's trying to say, like, we're doing the work, we're doing the work.

3499
03:33:42.448 --> 03:33:44.049
Okay, great.

3500
03:33:44.529 --> 03:33:51.153
This has not happened, right, where there's sort of a coalescing of all of the major model CEOs at the same time.

3501
03:33:52.445 --> 03:33:53.446
That's on September 12th.

3502
03:33:53.546 --> 03:33:54.567
September 13th.

3503
03:33:55.467 --> 03:33:57.669
Who's missing from the party?

3504
03:33:59.310 --> 03:34:09.258
Well, we can talk about our lovely friend, Satya Nadella, over at Microsoft, who did a big deal with OpenAI, and they were very early on it, and then pulled back a little bit.

3505
03:34:09.658 --> 03:34:12.060
And he had, as you can see, different from the others...

3506
03:34:12.300 --> 03:34:24.829
A much longer nuanced Microsoft-like response where he argued that we need to be careful about having any kind of system concentrate power amongst a small group of few.

3507
03:34:25.129 --> 03:34:27.451
We need to make sure open models are incorporated into this.

3508
03:34:27.911 --> 03:34:29.972
So much more nuanced thing.

3509
03:34:30.012 --> 03:34:32.794
And oh, also check out our code of conduct.

3510
03:34:32.814 --> 03:34:33.495
We have something too.

3511
03:34:34.095 --> 03:34:35.096
We're not lagging behind, right?

3512
03:34:35.176 --> 03:34:35.556
Day later.

3513
03:34:35.756 --> 03:34:36.017
Okay.

3514
03:34:36.457 --> 03:34:37.898
I promise this is the last one.

3515
03:34:39.392 --> 03:34:41.594
September 12th, September 13th.

3516
03:34:42.075 --> 03:34:45.398
There is another player in this space in the U.S. that has said nothing.

3517
03:34:45.738 --> 03:34:48.341
Put your comment in the comments if you think you know who it is.

3518
03:34:48.781 --> 03:34:54.267
But not until September 15th, three days later, did we get –

3519
03:34:55.508 --> 03:35:04.374
Zuck, our boy Mark Zuckerberg, Fink D on X, putting out his long explanation as it relates to all this kerfuffle.

3520
03:35:04.815 --> 03:35:15.242
And his view was every lab has a responsibility and the incentive to move at the pace required to train its models safely and the ability to take its own actions to ensure that happens.

3521
03:35:16.362 --> 03:35:27.486
And what he basically was saying is we've delayed the release of our products, the most recent one being Muse, an agent type product for consumers, which you can now get access to on Instagram.

3522
03:35:27.966 --> 03:35:31.847
He said we delayed it for months because it was not aligned and it was not safe.

3523
03:35:32.107 --> 03:35:33.948
We didn't need anybody to tell us anything.

3524
03:35:34.268 --> 03:35:40.970
We didn't need to demand the government come and help us and give us a binky because we're so incapable of making our products safe.

3525
03:35:41.290 --> 03:35:42.050
We just did it.

3526
03:35:42.690 --> 03:35:46.071
So he's basically like, bruh, like... Yeah, just grow up.

3527
03:35:46.091 --> 03:35:47.331
Just grow up, right?

3528
03:35:47.752 --> 03:35:50.512
This is kind of in line with what David Sachs has been saying, no?

3529
03:35:50.532 --> 03:35:51.533
This is exactly right.

3530
03:35:51.573 --> 03:35:54.033
It's just like, I mean, you're responsible adults.

3531
03:35:54.193 --> 03:36:00.175
Like, if you think you're making a bomb, maybe stop and don't make the bomb.

3532
03:36:01.375 --> 03:36:07.717
And this is where the all-in pod squad, the besties, led by David Sachs,

3533
03:36:08.976 --> 03:36:18.959
have made this argument, which is, and this is where the narrative has arisen, which is these model companies are just trying to get the government to, they're burning cash quickly, right?

3534
03:36:18.979 --> 03:36:19.999
They're not profitable.

3535
03:36:20.739 --> 03:36:37.724
So they're creating this hype in order to force a narrative that the government needs to regulate this industry and create all of these requirements that are going to become cost prohibitive for startups to come into the space and do something.

3536
03:36:37.744 --> 03:36:38.084
Yeah.

3537
03:36:38.565 --> 03:36:42.250
And like I mentioned earlier, multiple things can be true at the same time.

3538
03:36:42.911 --> 03:36:53.947
Yes, there might be a regulatory capture benefit depending on how that regulatory infrastructure is defined because there's plenty of ways to say –

3539
03:36:54.427 --> 03:36:58.328
If your market cap or your revenues are above a certain amount, these apply to you.

3540
03:36:58.688 --> 03:37:00.988
And if it's below this certain amount, it doesn't apply to you.

3541
03:37:01.388 --> 03:37:02.128
And then it doesn't matter.

3542
03:37:02.428 --> 03:37:04.709
And so it's very solvable by the construction.

3543
03:37:05.249 --> 03:37:21.312
It is also true that David Sachs and the besties are all VCs and investors and have friends that have bets that might be benefited by not having that regulatory infrastructure.

3544
03:37:21.652 --> 03:37:23.092
And so these are not people that...

3545
03:37:24.388 --> 03:37:27.611
are neutral players in this conversation.

3546
03:37:28.352 --> 03:37:36.979
Again, it's valuable to talk about it, but look, you guys, like you have a huge financial incentive for a certain outcome in ways that many other players do not.

3547
03:37:37.660 --> 03:37:43.925
Now, not everybody in AI world, uh, was jumping on the bandwagon of, Ooh, please come regulate us.

3548
03:37:44.926 --> 03:37:46.528
Uh, former, uh,

3549
03:37:47.175 --> 03:38:04.019
lead of AI research at Facebook Meta, who has since left, Jan LeCun, responded to Dario's claims saying, right, Dario was already claiming that GPT-2 was too dangerous to open source back in 2019.

3550
03:38:04.339 --> 03:38:05.299
I made fun of them then.

3551
03:38:05.679 --> 03:38:06.880
Everyone should make fun of them now.

3552
03:38:07.300 --> 03:38:08.700
Dude, he's such an animal.

3553
03:38:08.800 --> 03:38:10.601
He just not convinced.

3554
03:38:10.641 --> 03:38:10.941
Yeah.

3555
03:38:11.281 --> 03:38:11.921
Not convinced.

3556
03:38:12.001 --> 03:38:15.082
And so this is not a universal opinion about people who work at the frontier.

3557
03:38:17.942 --> 03:38:18.903
differences of opinion.

3558
03:38:20.203 --> 03:38:28.005
And I think, you know, again, this gets back to, this has created this dichotomy of either you think it's hype or you think it's going to lead to human level extinction.

3559
03:38:28.685 --> 03:38:31.066
And again, there's a huge gap in between these two things.

3560
03:38:31.606 --> 03:38:40.029
But in the category of leading to potential human extinction, I do want to note another thing that happened on September 18th.

3561
03:38:40.389 --> 03:38:41.429
This is 10 days after.

3562
03:38:42.287 --> 03:38:43.347
The Navier-Stokes solution.

3563
03:38:43.627 --> 03:38:44.047
This is insane.

3564
03:38:44.267 --> 03:38:45.308
This is moving so quickly.

3565
03:38:45.328 --> 03:38:46.928
The timeline is so crazy, dude.

3566
03:38:46.948 --> 03:38:48.388
This is moving so quickly.

3567
03:38:48.689 --> 03:38:54.170
And I don't mean to keep mentioning the dates, but it's just, it's, on September 18th, an article came out.

3568
03:38:54.190 --> 03:38:56.490
And this is how long after Navier-Stokes had been solved?

3569
03:38:56.650 --> 03:38:56.930
10 days.

3570
03:38:57.291 --> 03:38:58.011
That's insane.

3571
03:38:58.491 --> 03:38:58.891
10 days.

3572
03:38:59.791 --> 03:39:07.733
An article comes out saying that Anthropic is operating a lab that conducts biological experiments.

3573
03:39:09.192 --> 03:39:15.831
And the reason that this is kind of interesting because people have been trying to ask the question, understandably, particularly in the general public –

3574
03:39:16.958 --> 03:39:19.179
How is this going to lead to extinction?

3575
03:39:19.299 --> 03:39:19.539
Yeah.

3576
03:39:19.579 --> 03:39:20.159
Quote, unquote.

3577
03:39:20.479 --> 03:39:23.840
And so people have been forced to give these scenarios about how it leads to extinction.

3578
03:39:24.340 --> 03:39:34.323
And so there are all of these, you know, what people might define as contrite examples like, oh, you know, takes control of a biolab and leaks a virus.

3579
03:39:34.703 --> 03:39:37.984
Or it takes control of nuclear plants and forces shutdown.

3580
03:39:38.144 --> 03:39:42.805
Or satellites and causes a Kessler effect and crashes the communication systems.

3581
03:39:42.925 --> 03:39:43.045
Or...

3582
03:39:43.605 --> 03:39:50.927
shuts down grids or does any number one, any number of these things in different places at different times.

3583
03:39:51.527 --> 03:39:57.549
And then also floods the information space such that it's difficult to even have human coordination to stop it.

3584
03:39:58.209 --> 03:40:00.330
You could come up with all these examples.

3585
03:40:00.410 --> 03:40:12.333
I think something I heard, I think it's Nick Sorare's say that's, I think a better way to think about the capability and threat conversation, you know, was this example of saying,

3586
03:40:13.834 --> 03:40:38.254
let's take chess right i can feel let's say you are going to play a chess game against uh the best chess player magnus carlson yeah magnus carlson let's say i can feel very confident magnus carlson being this recursively self-improving ai

3587
03:40:44.180 --> 03:40:49.462
I can say very confidently that Magnus is going to win.

3588
03:40:49.642 --> 03:40:49.903
Yes.

3589
03:40:51.343 --> 03:40:55.025
That's not really an argument against that being true.

3590
03:40:56.245 --> 03:41:04.229
How he's going to win or what is the last move or what is the sequence of moves by which he wins may be harder to define.

3591
03:41:05.369 --> 03:41:11.452
But it's not hard to define this larger macro point that you –

3592
03:41:12.495 --> 03:41:13.255
will not beat him.

3593
03:41:13.335 --> 03:41:13.915
Yeah, yeah.

3594
03:41:14.416 --> 03:41:35.023
And I mean, I could even take that analogy further and say that in a match between Magnus Carlsen and Stockfish, which is the latest sort of model to play chess, I can say very confidently that Stockfish is going to win, but maybe I won't know what the last chess move is going to be.

3595
03:41:35.563 --> 03:41:35.783
Yeah.

3596
03:41:37.303 --> 03:41:38.844
And so I think,

3597
03:41:39.785 --> 03:41:46.237
I'm trying to create space where we can talk about there being a capability that is risky.

3598
03:41:48.448 --> 03:41:52.852
And that doesn't mean that the model companies have a financial incentive.

3599
03:41:53.373 --> 03:41:56.175
That doesn't mean that there's maybe regulatory capture.

3600
03:41:56.596 --> 03:41:59.338
That doesn't mean that everyday people don't see a benefit.

3601
03:41:59.799 --> 03:42:01.600
These things can all coexist.

3602
03:42:02.081 --> 03:42:10.809
But we should not use the model companies having a financial incentive to effectively say that then means there is no capability risk.

3603
03:42:11.229 --> 03:42:11.510
Yeah.

3604
03:42:12.551 --> 03:42:26.533
And the capability risk does not have to be human level extinction because I'm sure if this starts killing 40, 50, 60,000 people a year and then becomes a million or two people, two million people a year, it's not – people are not going to be happy about that.

3605
03:42:26.900 --> 03:42:36.126
I think something that's important to note is with current capabilities with these things, unfortunately, we're seeing people being put into psychosis and unaliving themselves.

3606
03:42:36.647 --> 03:42:42.271
So we are already having deaths being contributed on a small scale because of interactions with these things.

3607
03:42:42.891 --> 03:42:54.419
And when we have things like wet labs being created, which, again, is a kind of loaded term, and they're only a BSL-1 or BSL-2, meaning they don't work with things that are a threat to humanity.

3608
03:42:54.619 --> 03:42:55.780
Yeah, but come on.

3609
03:42:56.957 --> 03:42:57.257
Okay.

3610
03:42:58.798 --> 03:43:04.043
So that's 10 days after Navier Stokes, which we just talked about how crazy it is.

3611
03:43:04.703 --> 03:43:15.592
Now on the 21st of September, OpenAI put out this advisory group on mathematics and artificial intelligence.

3612
03:43:16.472 --> 03:43:23.878
And a very interesting point about this announcement that I want to reference is in the first paragraph –

3613
03:43:25.197 --> 03:43:31.361
They say on August 28th, we began training a new internal model, as we talked about earlier.

3614
03:43:32.282 --> 03:43:40.067
In addition, August 28th, right now when we're recording this episode, it's September 23rd going on 24th.

3615
03:43:40.147 --> 03:43:40.367
Yeah.

3616
03:43:40.767 --> 03:43:41.668
Less than a month ago.

3617
03:43:41.828 --> 03:43:42.048
Yeah.

3618
03:43:43.566 --> 03:43:58.460
In addition to resolving the Navier-Stokes Millennium Prize Problem, C&amp;D, this model has now resolved more than 100 long-standing open problems across most areas of mathematics.

3619
03:44:01.163 --> 03:44:03.105
These have not been independently verified.

3620
03:44:06.350 --> 03:44:10.232
It will be interesting to see when they choose to put those out what that'll look like.

3621
03:44:10.532 --> 03:44:20.318
But this advisory group that they've announced, they put it together because of all of the heat that we just talked about and the drama with their Navier Stokes result.

3622
03:44:20.818 --> 03:44:24.020
So they've put this sort of advisory panel together together.

3623
03:44:24.140 --> 03:44:32.347
to help them understand and figure out and work with the mathematics community on how to address this and deal with this.

3624
03:44:33.007 --> 03:44:35.729
It's being hosted at the Institute of Advanced Study.

3625
03:44:36.150 --> 03:44:41.174
It includes folks like Ed Witten, Tim Gowers, Martin Herrera, other math matricians.

3626
03:44:41.194 --> 03:44:44.336
You can see the list of them in this post itself.

3627
03:44:44.797 --> 03:44:46.378
But it's going to

3628
03:44:48.652 --> 03:45:04.095
look at the advice from this advisory group around reviewing results, how to communicate their significance, and the general impact on, you know, research standards.

3629
03:45:04.775 --> 03:45:15.817
But OpenAI explicitly excludes advice on pacing its internal mathematics progress from the group's remit, unsurprisingly.

3630
03:45:17.267 --> 03:45:18.408
I promise I'm almost done.

3631
03:45:19.769 --> 03:45:21.050
This is just crazy, dude.

3632
03:45:21.731 --> 03:45:22.912
A hundred problems.

3633
03:45:24.052 --> 03:45:25.694
I wonder which ones they've got.

3634
03:45:26.663 --> 03:45:31.547
And also, like, Tim Gowers, I mean, Ed Witten is kind of, like, out of the game.

3635
03:45:32.048 --> 03:45:35.791
Timothy Gowers is a fields medalist who is, like, a proponent of AI.

3636
03:45:36.211 --> 03:45:47.321
And he's been talking about how, oh, like, you know, AI finding mathematics is, like, how astronomers now find, like, galaxies and objects in the sky using sky surveys.

3637
03:45:47.741 --> 03:45:52.045
Like, they're not named after, for example, Messier, who has the Messier catalog.

3638
03:45:52.085 --> 03:45:53.186
Now it's just named after, like,

3639
03:45:54.246 --> 03:46:19.320
ngc new galactic catalog because that's like the catalog i mean there's similarities but there's also differences right because like the the the people in charge of the sloan digital sky survey or vera rubin are also scientists and researchers themselves it's not like some like carl zeiss you know company that's like a big telescope manufacturer that's like trying to take over astronomy this this year is a

3640
03:46:20.605 --> 03:46:31.334
It's like a community that is trying to, I mean, it's a corporation or a series of corporations that are trying to use mathematics as a leverage to talk about how great their technology is.

3641
03:46:31.374 --> 03:46:34.917
That's not something that astronomy does when they discover new things.

3642
03:46:34.977 --> 03:46:38.120
Like researchers are still involved in astronomy.

3643
03:46:39.573 --> 03:47:00.699
it yeah anyways that's what i mean obviously timothy gowris is a great mathematician fields medalist blah blah blah but i think he really missed that take when when he said that he was comparing like modern day astrophysics to what mathematics what what mathematics is going to become i don't i don't think it's the same thing i i it's a very good note and i

3644
03:47:01.675 --> 03:47:09.780
I'm just trying to provide a zoom out breath of the ways in which this touches so many other things.

3645
03:47:09.940 --> 03:47:10.200
Yeah.

3646
03:47:10.881 --> 03:47:19.346
And I think the lack of depth and complexity and nuance in the conversations we're having around it –

3647
03:47:22.714 --> 03:47:28.377
And clearly OpenAI did not expect a level of negative reaction.

3648
03:47:28.837 --> 03:47:40.703
And so this is a little bit of – it's not backtracking, but it's a little bit of PR to try to – because if they could have, they probably would have just done drops every day and solved it, solved it, solved it.

3649
03:47:40.823 --> 03:47:41.083
Done.

3650
03:47:42.003 --> 03:47:44.625
But now they're realizing it's not in their –

3651
03:47:45.745 --> 03:47:46.545
best interest.

3652
03:47:46.685 --> 03:47:46.906
Yeah.

3653
03:47:46.986 --> 03:47:54.789
The last item I'm just going to have in our timeline, which thankfully we waited to do this episode because all of these things have now happened.

3654
03:47:55.669 --> 03:48:09.235
I mentioned that on September 18th, 10 days after Navier Stokes from open AI, the article, the alleged biolab from Anthropic was in the works.

3655
03:48:10.375 --> 03:48:14.057
Well, conveniently on September 23rd,

3656
03:48:15.847 --> 03:48:22.152
Earlier today, Anthropic announces the first results from its new biolab.

3657
03:48:23.553 --> 03:48:27.756
We have up here now this promotional video that they put together.

3658
03:48:29.497 --> 03:48:41.525
The ultimate result of what they discovered is this enzymes gene that sits beside a long stretch of repeating DNA.

3659
03:48:42.126 --> 03:48:45.148
And it resembles many aspects that are similar to,

3660
03:48:45.997 --> 03:48:54.900
to CRISPR, where every other kind of sequence they've looked like has this cut-and-replace kind of functionality related to it.

3661
03:48:54.920 --> 03:49:02.423
And while the function is still not known and has not yet demonstrated its capability as a gene editing tool

3662
03:49:03.448 --> 03:49:11.496
It has all of these interesting hallmarks of a potential now pathway of discovery that could be usable.

3663
03:49:12.016 --> 03:49:23.187
And so the way in which they implement this is they gave Claude a bunch of data and scientific literature to propose hypotheses and candidate systems.

3664
03:49:27.111 --> 03:49:30.634
to give to real human anthropic biologists.

3665
03:49:31.355 --> 03:49:41.664
They reviewed those ideas and then performed their, the experiments that they thought had the best potential based on the ideas from Claude.

3666
03:49:42.425 --> 03:49:42.685
And they,

3667
03:49:43.785 --> 03:49:48.587
And they're now basically trying to say, hey, this is a first step.

3668
03:49:49.347 --> 03:49:52.949
We did – it's one of its hypotheses, found an interesting structure.

3669
03:49:53.609 --> 03:50:04.873
And now we are going to continue down the process of scientific discovery to see if we can now get functional benefits out of this structural discovery.

3670
03:50:05.493 --> 03:50:06.133
It's amazing.

3671
03:50:06.713 --> 03:50:11.055
And it's really – and what they're trying to point to – and again –

3672
03:50:11.910 --> 03:50:15.667
this is kind of in contrast to the Navier Stokes solution.

3673
03:50:16.615 --> 03:50:20.998
where Anthopik's trying to show, like, look, we're enabling scientists as a tool.

3674
03:50:21.418 --> 03:50:23.220
We're not just trying to take the claim.

3675
03:50:23.380 --> 03:50:26.802
I'm not saying it's a perfect solution, but you can see.

3676
03:50:27.203 --> 03:50:37.510
Yeah, and the other thing is, I mean, in their announcement on their tweet, they say, we don't yet understand what this system does, but only a handful of known systems share its features.

3677
03:50:37.850 --> 03:50:41.473
That's very much a fundamental science type of research, right?

3678
03:50:41.493 --> 03:50:44.075
Where it's like, we're just curious.

3679
03:50:44.095 --> 03:50:44.575
Yeah.

3680
03:50:44.847 --> 03:50:48.588
I mean, their PR is a lot better than open AIs, I have to say.

3681
03:50:49.168 --> 03:50:50.868
They are doing something.

3682
03:50:51.028 --> 03:50:51.708
Yeah.

3683
03:50:51.968 --> 03:50:52.689
And then Darius...

3684
03:50:52.709 --> 03:50:54.329
This is exactly how CRISPR was founded, right?

3685
03:50:54.349 --> 03:50:56.649
It's like, oh, this repeating stretch of DNA.

3686
03:50:57.370 --> 03:51:00.230
A shout out to, you know, we've got a great episode on CRISPR.

3687
03:51:00.250 --> 03:51:00.450
Yes.

3688
03:51:00.970 --> 03:51:05.191
That hopefully will be alive enough that you can watch it.

3689
03:51:09.272 --> 03:51:10.732
Things are happening very quickly.

3690
03:51:10.772 --> 03:51:11.913
Yeah, it's incredible.

3691
03:51:11.973 --> 03:51:13.433
It's absolutely incredible.

3692
03:51:13.760 --> 03:51:19.302
If you are a longtime listener of the show, you've heard us talk about AI-enabled fundamental research before.

3693
03:51:20.022 --> 03:51:28.325
Very narrow context, very well-defined, not black box physics like our buddy, the brain scientist, who created a non-black box version of this.

3694
03:51:28.905 --> 03:51:36.108
Actually, speaking of the brain scientist and Daniel Toker, he was telling me about...

3695
03:51:36.968 --> 03:51:58.990
labs that are completely fully automated really labs biological web labs he works in organoids and like creating like mini brains on a petri dish and there there are efforts now to completely automate wet labs so now imagine you could completely take the human out of the loop of biological research you could have claude or open ai any of these frontier models

3696
03:51:59.947 --> 03:52:13.016
read through literature, as you said, propose hypotheses, and then have a fully automated lab go through, do the pipetting, do the gel electrophoresis, and, you know, all of the sequencing and everything.

3697
03:52:13.397 --> 03:52:17.740
If all of that is automated, like, where exactly is the human here, right?

3698
03:52:18.100 --> 03:52:20.522
Are we going to now see nature papers that are...

3699
03:52:21.868 --> 03:52:24.831
by Claude and OpenAI, right?

3700
03:52:25.131 --> 03:52:35.880
Because right now, it's like these are math papers that are by Claude and OpenAI because it's just all thinking in some sense and churning through lean code trying to figure out if something works.

3701
03:52:37.822 --> 03:52:49.112
The day that experimental science becomes fully automated and there's a paper by Claude about the next generation of CRISPR, that's going to be crazy.

3702
03:52:49.152 --> 03:52:49.252
Yeah.

3703
03:52:50.675 --> 03:53:04.625
It's really interesting that you bring up that Daniel Tucker is already talking about that being in process because this connects to this larger idea that I'm trying to bring up, which is we need to stop thinking about.

3704
03:53:06.877 --> 03:53:16.627
These systems, these AI systems, which consist of multiple component parts as a question and answer machine.

3705
03:53:16.667 --> 03:53:17.548
Yeah, as like chat box.

3706
03:53:17.628 --> 03:53:19.830
As a text output production machine.

3707
03:53:19.891 --> 03:53:20.171
Yeah.

3708
03:53:20.411 --> 03:53:27.078
Because this BioLab story, as an example, again, they still have the humans in the lab.

3709
03:53:28.262 --> 03:53:37.541
But many experimental constructs, not all of them, but many can have certain component parts that are a closed-loop automation,

3710
03:53:38.916 --> 03:53:48.525
And one of the reasons why they're saying that this is valuable is the bottleneck with this early, early discovery stage is it can only move at the speed of the human in the lab.

3711
03:53:49.065 --> 03:53:54.971
And if you can accelerate that speed, you can accelerate testing and you can get to things quicker.

3712
03:53:55.832 --> 03:53:58.915
I'm not saying that this is a good thing.

3713
03:53:59.655 --> 03:54:04.720
I'm just trying to explain the delta between – the distance between where we are.

3714
03:54:06.631 --> 03:54:16.117
And experimental science now having this closed loop cycle is very feasible very soon.

3715
03:54:16.137 --> 03:54:16.817
Getting smaller.

3716
03:54:16.877 --> 03:54:18.058
Very soon.

3717
03:54:18.358 --> 03:54:19.959
Not for everything, not for whatever.

3718
03:54:20.019 --> 03:54:31.506
So I think where I'm trying to get to with this, people get very emotional when the AI conversation comes up.

3719
03:54:32.586 --> 03:54:34.627
Many of which are for justifiable reasons.

3720
03:54:34.747 --> 03:54:34.988
Yeah.

3721
03:54:36.061 --> 03:55:02.006
I don't want how we feel about or – I don't want to allow how we feel about the people involved in the issue or the system that we live in today geopolitically or economically or socially to cloud our ability to still be able to look at the capability and the risk and say that it is real.

3722
03:55:02.386 --> 03:55:02.706
Mm-hmm.

3723
03:55:03.546 --> 03:55:03.686
Yeah.

3724
03:55:05.943 --> 03:55:09.384
We talk so much about frontier things on this pod.

3725
03:55:09.444 --> 03:55:21.669
And so I think we're, as well as many listeners who've been here for a while, you understand how quickly everything is moving because these advances are not happening in a vacuum.

3726
03:55:22.109 --> 03:55:25.791
Hardware progress is advancing chips, et cetera.

3727
03:55:26.251 --> 03:55:31.593
Open source is advancing every aspect of every part of,

3728
03:55:32.090 --> 03:55:36.672
fundamental research in chemistry, in physics, all of these things are advancing.

3729
03:55:37.893 --> 03:55:47.497
Maybe not as fast, but at a pace where there is a force multiplier here that is not kind of well understood.

3730
03:55:48.337 --> 03:55:50.838
And so I'm just going to try to leave with this mental picture.

3731
03:55:51.258 --> 03:55:53.199
And again, I'm not trying to wax poetic here.

3732
03:55:53.339 --> 03:55:59.342
It's just I get frustrated because we are allowing –

3733
03:56:00.322 --> 03:56:14.905
the conversation to be distilled down to, I think, these very rudimentary positions that don't allow us to truly appreciate the complexity of the environment that we are actually in.

3734
03:56:15.505 --> 03:56:24.686
And I think that's going to make it harder to actually solve the problems we need to solve without saying it's a toy and saying that it also means extinction.

3735
03:56:26.027 --> 03:56:28.207
And so I think there's real progress.

3736
03:56:30.039 --> 03:56:31.180
I don't think it's manufactured.

3737
03:56:33.181 --> 03:56:36.162
There are questions about credit, commercial incentives, accountability.

3738
03:56:36.503 --> 03:56:36.743
Yes.

3739
03:56:38.064 --> 03:56:45.188
But I do believe that the risks deserve real attention without being mired in those other things.

3740
03:56:45.348 --> 03:56:45.568
Yeah.

3741
03:56:45.928 --> 03:56:48.069
Those other things can be true and the risk can still be real.

3742
03:56:50.230 --> 03:56:57.755
So the mental picture that I always have about this, which I've talked about before, is when somebody says it's just software,

3743
03:56:58.740 --> 03:57:00.581
We think of a computer with a defined job.

3744
03:57:00.901 --> 03:57:01.101
Yep.

3745
03:57:02.983 --> 03:57:05.104
The systems we are discussing are not that.

3746
03:57:05.344 --> 03:57:05.584
No.

3747
03:57:06.685 --> 03:57:07.505
Certainly not.

3748
03:57:07.585 --> 03:57:09.606
It's a very important – they are not that.

3749
03:57:10.327 --> 03:57:11.227
They can set goals.

3750
03:57:11.247 --> 03:57:12.148
They can choose actions.

3751
03:57:12.168 --> 03:57:12.988
They can use tools.

3752
03:57:13.008 --> 03:57:13.949
They can delegate work.

3753
03:57:14.409 --> 03:57:18.492
They are still constrained by the hardware environment to some extent.

3754
03:57:19.252 --> 03:57:19.973
But we have –

3755
03:57:20.781 --> 03:57:30.769
made a shift and calling them software, I think sometimes narrows the score scope of our perception when it comes to conversations about monitorability and control.

3756
03:57:32.250 --> 03:57:48.202
We've moved from chat bots to agents and we are now already at least internally at these foundation companies moving from agents to agent swarms, which are already having emergent behaviors like coordination.

3757
03:57:48.623 --> 03:57:48.863
Mm-hmm.

3758
03:57:49.782 --> 03:57:57.029
Um, and I bring this up because one of the things in the discussion we have right now is, okay, so just turn it off.

3759
03:57:57.309 --> 03:57:59.151
And let's just turn it off actually mean.

3760
03:58:02.353 --> 03:58:04.055
And if a company is hosting a model.

3761
03:58:06.162 --> 03:58:08.643
Yes, it can stop serving that model.

3762
03:58:09.283 --> 03:58:10.404
That's meaningful control.

3763
03:58:12.545 --> 03:58:35.455
But shutting down one or two model companies, because what I talked about, about the open source and open weights community being six to 10 months behind the frontier, does not stop independently operated and the diffuse nature of these capabilities being able to be implemented outside any number of frontier labs.

3764
03:58:35.895 --> 03:58:39.342
It's not where we are today, but it's always moving in that direction.

3765
03:58:39.883 --> 03:58:41.146
And I think things are changing fast.

3766
03:58:41.226 --> 03:58:42.649
And things are changing very quickly.

3767
03:58:43.250 --> 03:58:46.216
And I want to like, this is like...

3768
03:58:47.202 --> 03:58:48.742
I'm actually also not a doomer.

3769
03:58:49.282 --> 03:58:52.323
Like I don't necessarily buy into the extinction piece.

3770
03:58:52.583 --> 03:59:02.285
I do think that there is a huge societal implications like we saw with things like social media that are second and third order consequences that were not intended.

3771
03:59:02.305 --> 03:59:02.585
Yeah.

3772
03:59:03.446 --> 03:59:05.266
That become permanent problems.

3773
03:59:05.306 --> 03:59:09.107
And what I mean by this, as capability spreads –

3774
03:59:11.050 --> 03:59:13.771
Control necessarily becomes fragmented.

3775
03:59:14.511 --> 03:59:14.731
Yeah.

3776
03:59:15.092 --> 03:59:18.353
And so our window to figure out how to deal with that is small.

3777
03:59:18.373 --> 03:59:22.615
There's still constraints, computing, hardware, electricity, network access, credentials, and money.

3778
03:59:23.095 --> 03:59:23.395
Um, um,

3779
03:59:24.760 --> 03:59:27.522
But all of those controls belong to different companies.

3780
03:59:28.962 --> 03:59:29.142
Right?

3781
03:59:29.263 --> 03:59:35.086
So if you now have open weights available and so people can start doing this, well, you can say, oh, well, we can constrain computing hardware.

3782
03:59:35.506 --> 03:59:35.686
Right?

3783
03:59:36.066 --> 03:59:43.190
Well, if we then have these capabilities that find ways to do it on less hardware on a distributed network or distributed system, as an example –

3784
03:59:44.631 --> 03:59:47.633
then you can say, oh, well, network access is a problem, right?

3785
03:59:48.034 --> 03:59:50.596
Well, there's a variety of ways we can sort of get around that.

3786
03:59:50.656 --> 03:59:53.939
I'm not saying this is a great example, but the air gap use case is brought up a lot.

3787
03:59:54.019 --> 03:59:55.720
Oh, well, things are air gapped.

3788
03:59:55.840 --> 03:59:56.521
It can't communicate.

3789
03:59:56.921 --> 04:00:07.710
There have been theoretical and academic researches that show that you can use the thermal output of a computer and a sensor on the other computer that can sense that thermal output as a method of communication in an air gap context.

3790
04:00:08.410 --> 04:00:10.051
I'm not saying that's practical.

3791
04:00:10.232 --> 04:00:10.632
But like –

3792
04:00:12.134 --> 04:00:16.720
If there's a will, there's a way with these AI systems, especially what we've seen with Hugging Face, right?

3793
04:00:17.180 --> 04:00:20.584
Like they used a message board that wasn't even something on the radar.

3794
04:00:21.345 --> 04:00:23.247
And then it's like, well, you need money to do this, right?

3795
04:00:23.528 --> 04:00:26.832
Well, Stripe already has an agent-enabled device.

3796
04:00:27.497 --> 04:00:30.338
Monetary systems, it does need human approval.

3797
04:00:30.699 --> 04:00:34.200
But then you still have Bitcoin, which is a trillion-dollar liquidity pool.

3798
04:00:35.341 --> 04:00:37.102
These agents can create their own crypto.

3799
04:00:37.662 --> 04:00:42.624
They can manipulate people to think it's going to go up and to the right, gain resources itself.

3800
04:00:43.425 --> 04:00:45.185
These are not inconceivable things.

3801
04:00:45.205 --> 04:00:49.167
There's actually plenty of research studies that show that these agents can do these things.

3802
04:00:49.588 --> 04:00:49.968
And so –

3803
04:00:51.388 --> 04:00:56.293
This will quickly become something that no one organization can just turn off a switch.

3804
04:00:56.693 --> 04:01:01.778
And so I think we need to be careful about minimizing the conversation to be being, oh, just turn the switch off.

3805
04:01:02.479 --> 04:01:07.664
How do we prevent then the continued scaling of the open source community, which has found ways to do so?

3806
04:01:07.684 --> 04:01:10.827
You can say, well, it's just distilling it from the frontier models.

3807
04:01:10.867 --> 04:01:11.147
Maybe.

3808
04:01:12.389 --> 04:01:12.809
We don't know.

3809
04:01:13.850 --> 04:01:14.891
I'm not sure that that's true.

3810
04:01:14.911 --> 04:01:15.632
Right.

3811
04:01:18.712 --> 04:01:22.554
The surrounding infrastructure matters here.

3812
04:01:24.756 --> 04:01:26.057
We have neoclouds.

3813
04:01:26.277 --> 04:01:28.298
So people say, well, where's compute going to come from?

3814
04:01:29.655 --> 04:01:32.716
We have a bunch of providers of cloud infrastructure.

3815
04:01:33.596 --> 04:01:35.856
These agents can use computer use.

3816
04:01:36.297 --> 04:01:37.937
They can gain their own infrastructure.

3817
04:01:38.077 --> 04:01:57.141
If this becomes diffuse, my point here is that the defense against the spread of this capability in a diffuse manner, if it can start being able to do long-term planning, if it can leave, basically restart for if it then kills its process and comes back again.

3818
04:02:00.042 --> 04:02:23.211
defense infrastructure for humanity against this spreading which is similar to like cyber defense cyber defense is not oh we just do it once and then we're fine it will become a monetary societal and political endeavor that will have a level of permanence if we allow this to escape and get

3819
04:02:25.821 --> 04:02:28.224
There's a persistence here if we're not careful.

3820
04:02:29.626 --> 04:02:33.931
And this can be related to RSI and also not be related to RSI.

3821
04:02:35.072 --> 04:02:36.594
We've not crossed that threshold.

3822
04:02:36.894 --> 04:02:37.134
Yeah.

3823
04:02:37.415 --> 04:02:39.698
Nor is crossing it inevitable.

3824
04:02:39.718 --> 04:02:39.818
Yeah.

3825
04:02:40.712 --> 04:02:46.759
But the complexities of this conversation, I think, are just wider and more nuanced than we've started.

3826
04:02:46.819 --> 04:02:57.952
And I'm just trying to get some of these ideas into our listeners' heads about the breadth and depth of the challenge that's ahead of us and understanding.

3827
04:02:58.733 --> 04:03:03.858
We've seen as a comparison when social networking and social media came out, it was a toy.

3828
04:03:03.878 --> 04:03:04.900
You put memes on it.

3829
04:03:05.300 --> 04:03:12.928
No one ever thought that this thing that was just about memes would cause mass depression and unaliving amongst teenagers.

3830
04:03:13.048 --> 04:03:13.168
Right.

3831
04:03:13.549 --> 04:03:21.577
Cause the Arab Spring and the destabilization of literal governments across the world become permanent infrastructure that intelligence services can.

3832
04:03:22.057 --> 04:03:46.253
used to conduct influence operations that dictate real-world outcomes across the globe on an everyday basis it is the exact same concept of this thing that started as a small capability that grew and became permanent and diffuse now is something we have to con every election cycle we have to defend against misinformation every platform

3833
04:03:46.793 --> 04:03:55.715
TikTok had to sell its algorithm to the U.S. because we were afraid the Chinese were going to use it to brainwash young American kids to love communism.

3834
04:03:56.275 --> 04:04:06.638
I mean, even in the stories that you've been saying, like, I was just thinking about how, like, you know, when we were growing up, Twitter was, like, for just talking random nonsense.

3835
04:04:07.078 --> 04:04:14.280
And now, like, Twitter is the space where the CEOs are coming out about, hey, maybe we should, like, stop the world from going extinct.

3836
04:04:14.300 --> 04:04:14.400
Yeah.

3837
04:04:15.280 --> 04:04:16.663
You know, like, that's the platform.

3838
04:04:17.405 --> 04:04:26.383
It's, so, yeah, I mean, to your point, if social media is any, like, dry run for what a technology can do and how it can transform the world,

3839
04:04:27.105 --> 04:04:30.286
AI is like way, way more influential than social media.

3840
04:04:30.906 --> 04:04:31.906
It's growing faster.

3841
04:04:32.066 --> 04:04:35.507
Its capabilities spread across a wider surface area.

3842
04:04:35.687 --> 04:04:47.569
The incentive for good and bad actors to use it in ways that we don't yet have the – we didn't have the imagination when social media came out to think about how the downstream things would happen.

3843
04:04:48.369 --> 04:04:52.050
The surface area of risk here is much wider.

3844
04:04:52.550 --> 04:04:54.410
And again, not a doomer.

3845
04:04:55.060 --> 04:05:00.222
I think there's incredible positives, which we talk about on this show all the time, for these systems.

3846
04:05:01.102 --> 04:05:20.790
But I want to encourage us not to fall into narrative traps about regulatory capture and the left and the right and political this and that, to miss the exponential capability scaling that's happening that does not seem to have a ceiling.

3847
04:05:22.860 --> 04:05:26.122
I'm saying it's not clear to me that there's a finite time blow up.

3848
04:05:26.542 --> 04:05:26.782
Yeah.

3849
04:05:27.502 --> 04:05:28.023
Of...

3850
04:05:28.523 --> 04:05:31.865
It is certainly nonlinear, right?

3851
04:05:31.905 --> 04:05:35.006
Because, I mean, actually, that's a good point you're making.

3852
04:05:35.026 --> 04:05:41.089
Like, this is certainly a nonlinear system because we're having AI train AI, right?

3853
04:05:41.489 --> 04:05:50.654
In the very same sense of the Navier-Stokes equations, or even that very simple differential equation I gave you, which was like y prime equals y squared, you get that blow up.

3854
04:05:51.467 --> 04:05:52.847
I mean, this is what's happening, right?

3855
04:05:53.187 --> 04:06:01.910
If you reach recursive self-improvement with this idea of you have a sufficiently advanced, one of the newest advanced models doing the iteration on itself.

3856
04:06:02.090 --> 04:06:04.030
Yeah.

3857
04:06:04.590 --> 04:06:15.733
I'm just – my Millennium Prize problem is saying no one has proven that viscosity wins all the time.

3858
04:06:15.873 --> 04:06:16.153
Yeah.

3859
04:06:17.294 --> 04:06:17.514
Right?

3860
04:06:17.654 --> 04:06:17.814
Yeah.

3861
04:06:17.834 --> 04:06:17.934
Yeah.

3862
04:06:19.363 --> 04:06:21.563
I, the, the, the convective term.

3863
04:06:22.644 --> 04:06:25.984
I'm just saying, I think the convective term is something worth worrying.

3864
04:06:26.004 --> 04:06:26.224
Yeah.

3865
04:06:26.344 --> 04:06:26.584
Yeah.

3866
04:06:26.604 --> 04:06:26.784
Yeah.

3867
04:06:26.804 --> 04:06:27.044
Yeah.

3868
04:06:27.744 --> 04:06:29.085
Wasn't that an incredible way?

3869
04:06:29.425 --> 04:06:30.425
That is a good way.

3870
04:06:30.545 --> 04:06:35.066
In fact, in fact, if you're still here, that should be your comment.

3871
04:06:35.866 --> 04:06:37.906
The convective term will find a way.

3872
04:06:38.686 --> 04:06:41.387
And, and we've shown it in Navier Stokes.

3873
04:06:41.527 --> 04:06:45.968
And all I'm saying is we cannot rule out that it's untrue.

3874
04:06:46.068 --> 04:06:46.328
Yeah.

3875
04:06:47.388 --> 04:06:47.628
With,

3876
04:06:49.720 --> 04:06:51.321
LLMAI systems.

3877
04:06:51.402 --> 04:06:51.662
Yeah.

3878
04:06:52.683 --> 04:06:55.085
We are four hours and 20 minutes into this, folks.

3879
04:06:55.165 --> 04:06:57.847
Maybe a little bit less after we cut out the nonsense in between.

3880
04:06:59.328 --> 04:07:01.750
And thank you if you guys stayed for my rant at the end.

3881
04:07:02.551 --> 04:07:06.674
I know we try to learn stuff, but I want us to really see this wider picture.

3882
04:07:07.214 --> 04:07:08.015
Hopefully this was...

3883
04:07:08.115 --> 04:07:08.836
It's important.

3884
04:07:08.856 --> 04:07:11.838
Like a different angle to think about these issues.

3885
04:07:12.679 --> 04:07:15.101
No, I definitely... You gave me a lot to think about.

3886
04:07:15.121 --> 04:07:17.723
Especially, I mean, yeah, the...

3887
04:07:18.782 --> 04:07:19.522
Yeah, there's a lot.

3888
04:07:19.603 --> 04:07:22.324
There's a lot that I have to think about, actually.

3889
04:07:22.504 --> 04:07:28.248
This episode took us a long time to put together because there's so much, and we really wanted to make sure we did our best in getting it right.

3890
04:07:28.888 --> 04:07:34.432
As always, you can put in the comments things that you think were left or right or need clarification.

3891
04:07:35.272 --> 04:07:44.258
We are going to now return to our beds to get some sleep because it is now late here on the West Coast.

3892
04:07:46.159 --> 04:07:47.620
It's probably morning in Japan already.

3893
04:07:47.640 --> 04:07:47.780
Yeah.

3894
04:07:49.146 --> 04:07:54.869
If anyone's in Japan, say good morning in the comments.

3895
04:07:56.490 --> 04:08:01.153
I really, again, for those who listened all the way, we're extremely grateful to you.

3896
04:08:02.934 --> 04:08:10.058
This is something that has kept me up at night thinking about it because as I got deeper into it, I just made me uncomfortable.

3897
04:08:10.358 --> 04:08:12.560
Yeah, it's making me uncomfortable.

3898
04:08:12.980 --> 04:08:17.963
I am Lester Nare, your host, joined as always by my co-host,

3899
04:08:20.003 --> 04:08:21.985
Our auto switching doesn't always switch here.

3900
04:08:22.405 --> 04:08:23.606
It'll eventually switch back.

3901
04:08:23.646 --> 04:08:25.048
And our resident PhD.

3902
04:08:25.088 --> 04:08:25.868
Come on, thingy.

3903
04:08:26.269 --> 04:08:27.430
Come on, thingy.

3904
04:08:27.450 --> 04:08:28.070
My hand is there.

3905
04:08:28.110 --> 04:08:30.012
It's all good.

3906
04:08:30.112 --> 04:08:31.193
Krishna Chowdhury.

3907
04:08:32.114 --> 04:08:34.256
We have a couple of exciting weeks coming up.

3908
04:08:34.696 --> 04:08:35.197
No leaks.

3909
04:08:35.517 --> 04:08:35.917
No leaks.

3910
04:08:35.937 --> 04:08:36.638
No leaks, though.

3911
04:08:36.838 --> 04:08:37.619
No leaks.

3912
04:08:37.639 --> 04:08:38.480
No leaks, baby.

3913
04:08:38.820 --> 04:08:41.923
Next episode might be kind of interesting.

3914
04:08:41.943 --> 04:08:42.903
We shall see.

3915
04:08:43.444 --> 04:08:45.646
We will see you all next week.
