Video Infoblog - Chaos: The Science of the Butterfly Effect

7 Dec 2019

Chaos theory means deterministic systems can be unpredictable. Thanks to LastPass for sponsoring this video. Click here to start using LastPass: https://ve42.co/VeLP Animations by Prof. Robert Ghrist: https://ve42.co/Ghrist Want to know more about chaos theory and non-linear dynamical systems? Check out: https://ve42.co/chaos-math Butterfly footage courtesy of Phil Torres and The Jungle Diaries: https://ve42.co/monarch Solar system, 3-body and printout animations by Jonny Hyman Some animations made with Universe Sandbox: https://universesandbox.com/ Special thanks to Prof. Mason Porter at UCLA who I interviewed for this video. I have long wanted to make a video about chaos, ever since reading James Gleick's fantastic book, Chaos. I hope this video gives an idea of phase space - a picture of dynamical systems in which each point completely represents the state of the system. For a pendulum, phase space is only 2-dimensional and you can get orbits (in the case of an undamped pendulum) or an inward spiral (in the case of a pendulum with friction). For the Lorenz equations we need three dimensions to show the phase space. The attractor you find for these equations is said to be strange and chaotic because there is no loop, only infinite curves that never intersect. This explains why the motion is so unpredictable - two different initial conditions that are very close together can end up arbitrarily far apart. Music from https://epidemicsound.com "The Longest Rest" "A Sound Foundation" "Seaweed".

Transcript

Intro

0:00

Part of this video is sponsored by LastPass.

0:02

More about last pass at the end of the show.

0:04

The butter fly effect is the idea that the tiny causes, like a flay of a butter fly's wings in Brazil,

0:10

can have huge effects, like setting off a tornado in Texas

0:14

Now that idea comes straight from the title of a scientific paper published nearly 50 years ago

0:20

and perhaps more than any other recent scientific concept, it has captured the public imagination

0:25

I mean on IMDB there is not one but 61

0:28

different movies, TV episodes, and short films with 'butterfly effect' in the title

0:33

not to mention prominent references in movies like Jurassic Park, or in songs, books, and memes.

0:39

Oh the memes

0:41

in pop culture the butterfly effect has come to mean

0:43

that even tiny, seemingly insignificant choices you make can have huge consequences later on in your life

0:50

and I think the reason people are so fascinated by the butterfly effect is because it gets at a fundamental question

0:56

Which is, how well can we predict the future?

1:00

Now the goal of this video is to answer that question by examining the science behind the butterfly effect

1:06

so if you go back to the late 1600s, after Isaac Newton had come up with his laws of motion and universal gravitation,

1:13

everything seemed predictable.

1:15

I mean we could explain the motions of all the planets and moons,

1:18

we could predict eclipses and the appearances of comets with pinpoint accuracy centuries in advance

1:23

French physicist Pierre-Simon Laplace summed it up in a famous thought experiment:

1:28

he imagined a super-intelligent being, now called Laplace's demon,

1:32

that knew everything about the current state of the universe:

1:35

the positions and momenta of all the particles and how they interact

1:40

if this intellect were vast enough to submit the data to analysis, he concluded,

1:45

then the future, just like the past, would be present before its eyes

1:50

This is total determinism: the view that the future is already fixed,

1:55

We just have to wait for it to manifest itself

1:58

I think if you've studied a bit of physics, this is the natural viewpoint to come away with

2:02

I mean sure there's Heisenberg's uncertainty principle from quantum mechanics,

2:06

but that's on the scale of atoms;

2:08

Pretty insignificant on the scale of people.

2:11

Virtually all the problems I studied were ones that could be solved analytically

2:15

like the motion of planets, or falling objects, or pendulums

2:19

and speaking of pendulums I want to look at a case of a simple pendulum here

2:24

to introduce an important representation of dynamical systems, which is phase space

Phase Space

2:30

so some people may be familiar with position-time or velocity-time graphs

2:34

but what if we wanted to make a 2d plot that represents every possible state of the pendulum?

2:38

Every possible thing it could do in one graph

2:41

well on the x-axis we can plot the angle of the pendulum,

2:44

and on the y-axis its velocity.

2:47

And this is what's called phase space.

2:49

If the pendulum has friction it will eventually slow down and stop

2:53

and this is shown in phase space by the inward spiral --

2:56

the pendulum swings slower and less far each time

3:00

and it doesn't really matter what the initial conditions are,

3:03

we know that the final state will be the pendulum at rest hanging straight down

3:07

and from the graph it looks like the system is attracted to the origin, that one fixed point

3:12

so this is called a fixed point attractor

3:15

now if the pendulum doesn't lose energy, well it swings back and forth the same way each time

3:20

and in phase space we get a loop

3:22

the pendulum is going fastest at the bottom but the swing is in opposite directions as it goes back and forth

3:28

the closed loop tells us the motion is periodic and predictable

3:32

anytime you see an image like this in phase space,

3:35

you know that this system regularly repeats

3:38

we can swing the pendulum with different amplitudes,

3:40

but the picture in phase space is very similar, just a different sized loop

3:44

now an important thing to note is that the curves never cross in phase space

3:49

and that's because each point uniquely identifies the complete state of the system

3:53

and that state has only one future

3:56

so once you've defined the initial state, the entire future is determined

4:00

now the pendulum can be well understood using Newtonian physics,

4:04

but Newton himself was aware of problems that did not submit to his equations so easily,

4:09

particularly the three-body problem.

4:11

so calculating the motion of the Earth around the Sun was simple enough with just those two bodies

4:16

but add in one more, say the moon,

4:18

and it became virtually impossible

4:20

Newton told his friend Haley that the theory of the motions of the moon made his head ache,

4:25

and kept him awake so often that he would think of it no more

4:29

the problem, as would become clear to Henri Poincaré two hundred years later,

4:33

was that there was no simple solution to the three-body problem

4:37

Poincaré had glimpsed what later became known as chaos.

Chaos

4:46

Chaos really came into focus in the 1960s,

4:49

when meteorologist Ed Lorenz tried to make a basic computer simulation of the Earth's atmosphere

4:54

he had 12 equations and 12 variables, things like temperature, pressure, humidity and so on

4:59

and the computer would print out each time step as a row of 12 numbers

5:03

so you could watch how they evolved over time

5:06

now the breakthrough came when Lorenz wanted to redo a run

5:09

but as a shortcut he entered the numbers from halfway through a previous printout

5:14

and then he set the computer calculating

5:16

he went off to get some coffee, and when he came back and saw the results,

5:20

Lorenz was stunned.

5:21

The new run followed the old one for a short while but then it diverged

5:26

and pretty soon it was describing a totally different state of the atmosphere

5:29

I mean totally different weather

5:31

Lorenz's first thought, of course, was that the computer had broken

5:34

Maybe a vacuum tube had blown.

5:36

But none had.

5:38

The real reason for the difference came down to the fact that printer rounded to three decimal places

5:42

whereas the computer calculated with six

5:45

So when he entered those initial conditions,

5:47

the difference of less than one part in a thousand

5:50

created totally different weather just a short time into the future

5:53

now Lorenz tried simplifying his equations and then simplifying them some more,

5:57

down to just three equations and three variables

6:00

which represented a toy model of convection:

6:03

essentially a 2d slice of the atmosphere heated at the bottom and cooled at the top

6:07

but again, he got the same type of behavior:

6:10

if he changed the numbers just a tiny bit, results diverged dramatically.

6:15

Lorenz's system displayed what's become known as sensitive dependence on initial conditions,

Sensitive Dependence

6:20

which is the hallmark of chaos

6:22

now since Lorenz was working with three variables, we can plot the phase space of his system in three dimensions

6:29

We can pick any point as our initial state and watch how it evolves.

6:33

Does our point move toward a fixed attractor?

6:37

Or a repeating loop?

6:39

It doesn't seem to

6:40

In truth, our system will never revisit the same exact state again.

6:46

Here I actually started with three closely spaced initial states,

6:50

and they've been evolving together so far, but now they're starting to diverge

6:55

From being arbitrarily close together, they end up on totally different trajectories.

7:00

This is sensitive dependence on initial conditions in action.

7:04

Now I should point out that there is nothing random at all about this system of equations.

7:09

It's completely deterministic, just like the pendulum

7:12

so if you could input exactly the same initial conditions

7:16

you would get exactly the same result

7:19

the problem is, unlike the pendulum, this system is chaotic

7:23

so any difference in initial conditions, no matter how tiny,

7:27

will be amplified to a totally different final state

7:31

It seems like a paradox, but this system is both deterministic and unpredictable

7:37

because in practice, you could never know the initial conditions with perfect accuracy,

7:42

and I'm talking infinite decimal places.

7:45

But the result suggests why even today with huge supercomputers,

7:49

it's so hard to forecast the weather more than a week in advance

7:52

In fact, studies have shown that by the eighth day of a long-range forecast,

7:57

the prediction is less accurate than if you just took the historical average conditions for that day

8:03

and knowing about chaos, meteorologists no longer make just a single forecast

8:08

instead they make ensemble forecasts,

8:11

varying initial conditions and model parameters

8:13

to create a set of predictions.

Chaos Everywhere

8:16

Now far from being the exception to the rule, chaotic systems have been turning up everywhere.

8:21

The double pendulum, just two simple pendulums connected together, is chaotic

8:26

here two double pendulums have been released simultaneously

8:29

with almost the same initial conditions

8:32

but no matter how hard you try,

8:35

you could never release a double pendulum and make it behave the same way twice.

8:39

its motion will forever be unpredictable

8:43

you might think that chaos always requires a lot of energy or irregular motions,

8:48

but this system of five fidgets spinners with repelling magnets in each of their arms is chaotic too

8:55

At first glance the system seems to repeat regularly,

8:59

but if you watch more closely, you'll notice some strange motions

9:03

a spinner suddenly flips the other way

9:06

Even our solar system is not predictable

9:08

a study simulating our solar system for a hundred million years into the future

9:13

found its behavior as a whole to be chaotic

9:16

with a characteristic time of about four million years

9:19

that means within say 10 or 15 million years,

9:23

some planets or moons may have collided or been flung out of the solar system entirely.

9:29

The very system we think of as the model of order,

9:32

is unpredictable on even modest timescales

9:36

So how well can we predict the future?

9:38

Not very well at all at least when it comes to chaotic systems

9:41

The further into the future you try to predict the harder it becomes

9:46

and past a certain point, predictions are no better than guesses.

9:50

The same is true when looking into the past of chaotic systems and trying to identify initial causes

9:56

I think of it kind of like a fog that sets in the further we try to look into the future or into the past

10:02

Chaos puts fundamental limits on what we can know about the future of systems

10:06

and what we can say about their past

10:09

But there is a silver lining

10:11

Let's look again at the phase space of Lorenz's equations

10:15

If we start with a whole bunch of different initial conditions and watch them evolve,

10:19

initially the motion is messy.

10:21

But soon all the points have moved towards or onto an object

10:25

the object, coincidentally, looks a bit like a butterfly.

10:29

it is the attractor

10:30

For a large range of initial conditions, the system evolves into a state on this attractor

10:36

Now remember: all the paths traced out here never cross and they never connect to form a loop,

10:41

If they did then they would continue on that loop forever and the behavior would be periodic and predictable

10:47

so each path here is actually an infinite curve in a finite space.

10:52

But how is that possible?

10:54

Fractals. But that's a story for another video

10:56

this particular attractor is called the Lorenz attractor,

10:59

Probably the most famous example of a chaotic attractor

11:02

though many others have been found for other systems of equations

11:06

now if people have heard anything about the butterfly effect,

11:08

it's usually about how tiny causes make the future unpredictable

11:12

but the science behind the butterfly effect also reveals a deep and beautiful structure underlying the dynamics

11:19

One that can provide useful insights into the behavior of a system

11:23

So you can't predict how any individual state will evolve,

11:27

but you can say how a collection of states evolves

11:30

and, at least in the case of Lorenz's equations,

11:33

they take the shape of a butterfly

LastPass

11:39

Hey this part of the video is sponsored by LastPass,

11:42

the password manager with unlimited password storage and free cross-device sync

11:46

before I used a password manager, I've got to admit, I used the same password for a lot of different accounts

11:52

and I know that is incredibly dangerous because if even one of those sites got hacked,

11:56

then all of my important accounts would be exposed

11:59

Quite the butterfly effect.

12:01

LastPass auto generates strong passwords for you

12:05

so you have a different, indecipherable code for each website

12:09

and let's face it: if there's anything in your life that you want to be chaotic,

12:12

it is the characters in your passwords

12:15

the best part is because it autofills user names and passwords on websites

12:19

or on iOS or Android apps and mobile sites,

12:22

you never have to remember another password again. Not for the rest of your life

12:27

and that means no more writing down passwords, no more getting locked out of accounts,

12:31

no more password resets. You can use your brain for what it's meant to be doing.

12:35

and if you want extra features like advanced multi-factor authentication,

12:39

you can upgrade to LastPass premium

12:41

I got to say, it works like magic.

12:43

so put your passwords on autopilot with LastPass

12:46

Click the link below to find out more,

12:47

and thanks to LastPass for sponsoring this part of the video