StarTalk Radio - The Language of the Universe with Grant Sanderson (3blue1brown)
Episode Date: May 20, 2025Why can’t you divide by zero? Neil deGrasse Tyson and Chuck Nice discuss higher dimensions, dividing by zero, and math’s unsolved questions with math YouTuber Grant Sanderson (3blue1brown).NOTE: S...tarTalk+ Patrons can listen to this entire episode commercial-free here:https://startalkmedia.com/show/the-language-of-the-universe-with-grant-sanderson-3blue1brown/Thanks to our Patrons Nicolas Alcayaga, Ryan Harris, Ken Carter, Ryan, Marine Mike USMC, VARD, Mile Milkovski, Gideon Grimm Gaming, Shams.Shafiei, Ben Goldman, Zayed Ahmed, Matt Nash, Stardust Detective, Leanice, morgoth7, Mary O'Hara, David TIlley, Eddie, Adam Isbell-Thorp. Armen Danielyan, Tavi, Matthew S Goodman, Jeremy Brownstein, Eric Springer, Viggo Edvard Hoff, Katie, Kate Snyder, Jamelith, Stanislaw, Ringo Nixon, Barbara Rothstein, Mike Kerklin, Wenis, Ron Sonntag, Susan Brown, Anti alluvion, Basel Dadsi, LoveliestDreams, Jenrose81, Raymond, David Burr, Shadi Al Abani, Bromopar, Zachary Sherwood, VP, Southwest Virginia accountability, Georgina Satchell, Nathan Arroyo, Jason Williams, Spencer Bladow, Sankalp Shinde, John Parker, Edward Clausen Jr, William Duncanson, Mark, and Dalton Evans for supporting us this week. Subscribe to SiriusXM Podcasts+ to listen to new episodes of StarTalk Radio ad-free and a whole week early.Start a free trial now on Apple Podcasts or by visiting siriusxm.com/podcastsplus.
Transcript
Discussion (0)
I think we need to do more episodes that feature math.
Yes, with maybe a different co-host,
because I don't do math.
No, I love math.
You love math?
I do.
You realize how important it is.
Well, you know, it's, yes.
Without math, we don't know nothing in this universe.
Well, I'm still in that boat no matter what.
Coming up, StarTalk Cosmic Queries,
a little bit of math.
Welcome to Star Talk, your place in the universe where science and pop culture
collide. Star Talk begins right now.
This is Star Talk Cosmic Queries edition.
Got Chuck nice with me.
Chuck, how you doing man?
I'm doing great.
Yeah, yeah.
This one is going to be on math.
Oh, I was told there would be no math.
Nobody told you there was going to be no math.
That's right.
I know some math, but if we're going to have math as a subject, we've got to bring on the
big guns.
Right.
Especially the Mathiest of the Mathiest.
We've got with us, and not his first time on Star Talk, Grant Sanderson.
Grant, welcome back.
Hey, thanks for having me again.
It was fun the first time.
Let's see how it goes.
Excellent, excellent.
So you have sort of academic chops in math and in computer science
All right, those are related on some levels and
You took it to the road. I mean you took it to YouTube. Yeah with a highly followed channel
Three blue one brown. Mm-hmm. Can you get more cryptic than that? Yes
Sounds like a three-card Monty's
Can you get more cryptic than that? Yes, I sound uh, sounds like a three-card monties
How long have you had the channel man it's been around 10 years now which uh feels a little wild
Yeah, and how many followers followers you have? It's uh, it's it's around seven million. Um, I remember sometime recently across, you know, that's that's
Monumental in terms of accomplishment. No, no, I just had seven million people to follow you for math math That's what I'm saying. Exactly. If seven million people that gives us gives me hope for the future of civilization
Yeah, that gives me hope for those seven million people
In general more people like math and people suspect I think like it's a little bit
It's a little bit of an underdog
Everyone thinks people hate it, but I think there's everyone loves math
I just think that most people are
Intimidated by it and intimidate and and nobody wants to feel stupid. I like feeling dumb. That's why I'm on this show
What I'm always the dumbest person on this show. But that's why Grant exists, so that he can grow
the comfort zone that people feel.
Look at you.
When encountering this content.
Yes.
So let me pick up some broad deep topics in math.
Before we get to our Q and A, we hear occasionally
about problems in math.
And no, they're not talking about what's eight times seven.
No, they're talking about problems
that the deepest thinkers in the field
have attempted to solve over the centuries, right?
Is there like a book of the latest unsolved problems
in math and then that's what nerd geeks
should check out of the library first?
Yeah.
Yeah. I mean, there's been many. I mean, one of the most famous bodies of unsolved problems
in the year 2000, the Clay Math Institute, put out these seven problems that they offered a
$1 million prize for. And so these ones are kind of the celebrities among unsolved problems. They're
called like the Clay Millennium Math Problems.
They are typically very hard to even describe
what the problem is stating.
So, very classic.
You can't even describe the problem
of the problem you're supposed to solve.
Right, you gotta solve the problem,
but you can't even describe,
nobody can tell you what the problem is.
Now to me, sir, that's-
That's how you keep your million dollars.
Yes.
I was going to say, that sounds a lot like being married.
But what is the problem even that we're beginning to solve?
But there's also these celebrity unsolved problems that are in some sense less important,
but they're easier to state.
And as a result, they're a lot more fun to just engage with for the public.
But there's no prize money for those, right?
Like not in the explicit sense that there is a particular institute that like put this
check behind it, but like absolutely if you solved any of these problems you gain a certain
fame within the math world, probably you're doing it as an academic where it really bolsters
your career.
Like there's plenty, if you want financial rewards there's like plenty that would come
if you could solve one of these problems.
But of course that's not what most people care about.
Are they arise purely within math
or is there some scientific pumping
of what these problems are or engineering solutions?
Yeah, so I'll give you one that's purely within math
and then some that are more like come from the outside world.
And this is one, it's near and dear to me
because I remember when I was, I don't know,
maybe 11 or so, my dad from across the kitchen,
he like pipes over, he's like,
hey Grant, you know prime numbers?
I'm like, oh yeah, prime numbers,
like numbers that you can't divide into two smaller pieces,
like six equals two times three,
but seven you can't break it up into two smaller pieces.
He's like, yeah, do you think there's infinitely many
of them that are just two apart?
So like 11 and 13 are primes that are two apart,
or 29 and 31 are a pair of primes that are two apart.
And he was asking this
because he was reading some news article that mentioned,
hey, this is an unsolved problem.
His dad asked him this question when he was 11.
Oh, wow.
So either you were really smart or you had a really bad dad.
So I was like, go and think, I'm like, oh, like, is this true?
And eventually it didn't take too long for him to say like, oh, this was a thing I was reading about in this like science news magazine that like mentioned this is a problem.
No one in the world knows how to answer.
And that was fascinating to me.
Like, here's a question you can ask.
You teach kids about prime numbers.
Maybe not everyone remembers them when they grow older, but it's a common topic.
All you're asking is, hey, are there infinitely many
that are two apart?
Do they ever stop coming spaced out by two?
And you know that primes get a little sparser.
As you get much bigger, there's fewer of them.
But you might wonder, do they stop clumping up in that way?
You could ask this over two millennia ago.
We still don't know the answer.
And it cuts pretty deep to the understanding of primes
that we have and the lack of understanding
to be able to answer questions like this.
So that's pure math.
It's just a puzzle.
And you got one from science?
Yeah, that's great.
Yeah, I mean, so one that's,
I'm not gonna be able to state the exact nature
of this question, but I can give you
the high-level overview, which I know, Neil,
you're gonna know this, but there's a set of equations that describe
fluid flow. They're very famous, they're called the Navier-Stokes equations.
Yeah. And so if you're just modeling some fluid and you understand
certain aspects of its, you know, pressure and viscosity and things like
this, there's something, for example, you could tell the computer to try to run
forward a simulation. But the theoretical understanding of these equations is like we're really thin in some
respects, where for example, it's not entirely known if it would imply that like you get
an infinite concentration of energy at some point.
You don't like clearly that shouldn't happen.
You don't think that would happen in the physical world.
But the mathematical model being used in the pure landscape of what are called differential equations.
It's got these properties where people aren't sure whether it falls one way or another.
It's actually very, very hard to understand the specific type of differential equation.
Again, I won't phrase the specific nature of the question that's unsolved, but broadly
speaking, it's some basic questions about do these equations
behave in the way that you would hope they would behave?
No one actually knows, and this is clearly motivated by modeling physics and modeling
the world.
Now, you also have a category of problems that are unsolvable, yet you have people who
think that they solved it, right?
So you can prove that something is unsolvable, correct?
Everyone might have seen in school the quadratic equation.
So this is something where if you have an expression that looks like x squared plus
some constant times x plus some constant equals zero and you want to solve it, it's a systematic
way to do it.
This is an equation that comes up all the time for engineers, all the times in like
computer graphics programming, just solving equations like this left and right.
And so there's a formula.
A lot of kids memorize it in school.
It's called the quadratic formula.
This is a formula that has been known for a really long time.
Chuck, recite the quadratic formula.
Do you remember?
Here's the quadratic formula.
What you get for number 13.
What do you get?
Looking over the shoulder.
But, I mean, it was, I remember it getting drilled into us
on a level where it prevented me from appreciating
what it was actually doing.
It was just the memorizable formula.
It was like negative B plus or minus the square root
of B squared minus four AC over two A.
Did I get that right?
That's exactly right, yeah.
Okay.
Okay.
I just did that by rote,
not because I had a fricking feeling for it. You just had to know it so many times. Yeah, I got it that by rote, not because I had a freaking feeling for it.
You just had to know it so many times.
Yeah. I got it from eighth grade. It came in. Okay.
Here's something you should be thankful for. So there exists a formula to solve any cubic equation,
but our teachers never made us memorize it. It's much longer. Yeah. And it would be,
it takes a much longer song to do it. And if they think the quadratic formula is this thing that
would cause a rote engagement with math
instead of a substantive one,
forcing kids to memorize the cubic would be even worse.
But like pure mathematicians for a while were curious,
okay, hey, how far can we take this?
Can we have a formula that solves any equation
where the highest power of X is X to the fourth?
And you can, it's an even more monstrous formula.
If you tried to write it down in full,
it would just feel like an entire page,
this formula that solves degree four.
And for a long time, the natural question is,
okay, can we solve any degree five equation?
And between the 1500s, when this was kind of initially
being explored with the cubic and everything,
up to around the 1800s, people tried,
but no one could find a way to answer that.
Wait, are you telling me people had these thoughts in the 1500s while others were disemboweling
heretics?
Yeah.
These two things coexisted.
It's wild to contrast history of science.
Well, maybe they stopped by the 1500s.
The Renaissance was going on.
Yeah, exactly.
I'm thinking maybe the Dark Ages.
Well, you go to the right part of the world.
One person's contemplating cubics, you can definitely find another part of the world
where disembowellings happen.
Just skipping to the punch line here, and I'll tell a little bit more story behind it,
the answer ended up being it's literally impossible in the appropriate sense.
If you're trying to write a formula using the usual symbols that we do, plus, minus,
times, divide, maybe you allow for roots and things like that. If these are the
operations at your disposal, you will literally never be able to write down a formula for
degree five or higher. And two very young mathematicians were the ones to make the initial
steps in this.
It works up to fourth order polynomial, but not fifth.
And then suddenly five, it stops. Yeah, it stops working.
Wow.
And there's a deep reason why.
And the discovery of this was one of the things
that gave birth to a huge part of modern math,
which is called abstract algebra.
Like most people don't know that this exists,
that there was this revolution in math in the 19th century
that kind of changed the way that we think about math.
But the beginnings of it were born out of this question of trying to solve polynomial
equations and realizing, hey, maybe it's impossible once you get above a certain point.
What's the reason?
Why does the symmetry break between four and five?
Neil, I would love so much to be able to give you the pithy answer that's like, here's what's
true at four that's not true at five. And I have struggled for years to think is this something I could do
in even like a 50 minute video or something that like compellingly describes it to the
general public.
I want a full YouTube video on my desk Monday morning of you explaining this.
It would make me and a lot of other people very happy. Okay, I'm going to give an answer that will make no sense, but I promise that there's
some sense in which this is true.
It's something called Galois theory.
One of the young mathematicians who came up with the arguments that led to explaining
this, his name was Everest Galois, he died in a duel at the age of 20.
This is his big headline fact.
He knew he was going to lose that duel. And so the night before
the duel, he like does like a brain dump on the page. Oh, this is the story. This is the classic
story that we all tell in the lecture halls. That's insane. Yeah, okay. Wait, let the man explain.
Go ahead. No, no. So what you're saying is what everyone hears when they're learning physics and
learning math and what not. Oh, that's not true? He wrote down all the... So he did, he like was compiling stuff that he had tried to get published like three or four times before.
So the stuff, he had written the stuff down before. He had gotten it in front of very famous mathematicians like Fourier or Cauchy and Poisson.
Like they had seen the beginnings of his work before. So it wasn't like the first time he's ever jotting this down was in the like crazed pre-dual state But it's such a good story to tell it that way, right? It's so good to be like he knew he's gonna die
You know, so he's got to get these ideas out in some way
He's got this classic story behind him in terms of the dual and all that but the math that he was doing
One of the things that it showed is that it's impossible to solve equations degree five or higher. It also showed other impossibilities. There's this
classic problem about trisecting the angle using a straight edge and compass to take any angle and divide it into three pieces. His theories, his new math, can prove
that that's not possible. It's simply too hard to describe right here over a podcast,
but it gave birth to a whole field of math that is both central to math and also particle
physics these days.
Wow, cool man.
Well math is the language of the universe.
It certainly is.
We shouldn't be surprised if it's some spillage
into cosmic discovery.
Yeah, makes sense.
So can we bring questions your way from our audience?
Yeah, yeah, yeah, let's do it.
Here's the first question from Buck Rice who says,
why can't we divide by zero?
Why?
I love that.
That's a great question.
I've carried that with me my whole life.
Right.
Because the answer is, well, it's undefined.
And so my response is, well, then define it.
Right.
Get off your duff and define it!
Yeah, exactly, define it. And you know who does?
The mathematicians, actually. There are parts of math,
one in particular called projective geometry,
where one of the objects in there is essentially what we want to get at
by the idea of one divided by zero.
And the thought is, if you have a number line and you walk infinitely far
either to the left or to the right,
there's this unified point that you're approaching called the point at infinity.
You can do useful math by defining that.
That's kind of what you're getting at when you have this notion of one divided by zero.
But if you're not doing that and you want to say, why isn't it defined?
It depends on what you're doing with division.
If what you want to say with division is like, I have one cupcake and I'm dividing it among
three people, how many does each of them get?
Like a third of a cupcake.
If I have one cupcake and I'm dividing it among zero people,
it's like it's an incoherent question.
Like the cupcake's gotta go somewhere.
The fact that that question is incoherent
is maybe what we mean by saying that it's undefined.
See, but that's my point.
In practical terms, I have one cupcake
and I want to divide it between zero people,
we got to go back to the beginning of the statement.
I have one cupcake.
Okay?
And that's the answer.
The answer is I eat the cupcake.
Now it's defined.
Now it's defined.
All right. This is Star Talk with Nailed Graz Tyson.
This is Kira. Actually, I'm going to combine two questions in one because Kira and Gavin Bamber actually are similar, but I'm going to read both their questions successively so you can answer
them. Okay. Hi, this is Keir from Georgia in the US.
In your opinion, what is the most fascinating
unsolved mathematical problem in cosmology
that if understood could fundamentally change
how we view the universe?
Hold that, Gavin Bamber says,
hey, Gavin here from North Vancouver.
Please visit, Neil.
What's your favorite unsolved math question
and how would you illustrate it?
So one, fundamentally change the way we view the universe
and then part B, what is your personal favorite unsolvable?
You can't say one and then part B.
It's either one and two or A and B.
No, see what I am doing is a new kind of math.
Tell you what, I'm going to punt off part one to Neil here because I have some humility here.
I'm not sure what the most important mathematical problems of cosmology are.
Not much of a cosmology person, so I'm curious what you say.
We have singularity problems in the universe.
All of our equations tell us that the center of a black hole, nature is dividing by zero.
Okay, everything goes, the denominator goes to zero.
What happens to the value of everything else?
We say there's infinite density and infinite this
and it doesn't even make any sense.
No it doesn't.
So what we don't know but we suspect is that
that's a limit to the application of our theory
of the universe, not a limit to the application of our theory of the universe,
not a limit to the invocation of the math.
Right.
Because we're not the first to blame the math.
I'm just saying.
Gotcha.
Because math is badass and we're not, okay?
So we're going to take the blame first.
But it is true that certain discoveries in math
have led the discoveries in math have led
the discoveries in astrophysics.
We had no need for non-Euclidean geometry until we did.
And so this is the curvature of space time.
It's not flat and Euclidean geometry is flat.
And who came up with curved geometry?
So Riemann is the big one there.
So Riemann, and when was that, like 1800s sometime?
Yeah, it was in the 1800s, like maybe 50s, let's say.
Okay, alright, so 19th century,
we have the tools to think about curved geometry.
That was immediately uptook by the cosmologists to think about what could be
the geometry of the universe. They needed a way to talk about it. So that's all I got
here, but it's the math leading us, not us finding a math problem that's not solved.
Well, I mean, you bring up Riemann talking about non-Riemannian geometry, but he's also
the source of how I was going to answer the part B of that question for one of my favorite unsolved problems.
There's part one and then part B.
Yeah, there's part one and part B.
So, Riemann, you know, he did a lot of geometry stuff.
He also was one of the fathers for complex analysis, basically using complex numbers
to solve other problems within math.
And he had one paper on number theory.
So that's the, I described prime numbers earlier like this twin prime
conjecture. He has this one paper that he puts out. I think it's 1857 about prime numbers.
Otherwise he doesn't do any number theory and it completely changed the whole field
because he basically said, hey, here's this continuous function. It doesn't feel like
it's about primes that are all discrete. It's very like continuous. It's got complex numbers.
That makes it very different.
And if you understand this function,
you completely understand the primes.
These days we call it the Riemann zeta function
because he used the Greek letter zeta.
And he basically said, hey, we can really, really
well understand how the primes are distributed
if we understand something about this function.
And he put this conjecture up about where all the,
if you want to solve when this function equals zero, he didn't know how to solve it.
He had a guess for where those solutions are and this is called the Riemann hypothesis.
He was hypothesizing it.
It's one of those million dollar problems and it clearly, it's a very, very beautiful
question because it's kind of asking like if the prime numbers form a chord in a certain
sense because it studies them based on frequency information.
And nobody knows how to answer it, but the more you dig into this question, it paints
a really, really beautiful picture.
Is that your favorite unsolved problem?
I think it's my favorite.
Yeah.
Oh, wow.
Okay.
It sounded like it was too.
You make it sound very elegant as a…
You sound a little bit excited as you're talking about it.
Yes.
I love it.
Okay.
All right. This is Tony Isaacs and Tony says,
g'day, Astro Nail and Lord Nice.
This is Tony here from Melbourne, Australia.
Melbourne?
Melbourne.
He says, love the show.
Go on, Chuck, and do an RC accident.
No.
No.
No.
No.
Okay, that's somebody who's been listening
to the show for a while.
Listening too long.
Yeah, yeah, yeah.
No. It totally telegraphed me.
All right, he says I have a complex problem.
It's the complex number I.
It's the square root of negative one,
which doesn't exist,
but it is used in so much math
that predicts things accurately,
including quantum.
It fries my brain.
I hope Neil and Grant can help me out, thanks.
So Grant, I'm going to lead off by saying,
why did you label those numbers imaginary?
My God, it's the worst name in all of that.
Worst name ever.
Gauss proposed calling them lateral numbers,
which would have been, you know.
A little better.
Just a little better. And then you add another aspect to the imaginary number
to get the complex number.
Right.
And now you call it a complex?
These are words that are complete turnoffs.
And I blame you.
Well, I mean, if that's the case.
I blame your people.
I was going to say, if that's the case,
then we need to go all the way back to the beginning
because we call them math problems.
And who wants to deal with problems?
So many people struggle with this, right? Because you label it as imaginary. You start by pitching it by saying, you know, square roots of negatives don't exist, but pretend like they do and run
forward. And like, you could teach the whole topic completely differently, where you start off by
talking about processes that cycle and trying to model processes that cycle and using our normal
number systems for that.
And anything that has cyclic kind of behavior, there's a natural number system to try to
describe that.
Call that number system what you want.
Oh, so you can think of them as like clocks then.
Think of them as clock numbers, right?
Anytime you're doing clock stuff, these numbers are going to be great.
So we use the clock numbers.
Why are they relevant in quantum mechanics?
You got a bunch of waves, right?
There's things that are cycling and there's frequencies that are relevant.
You've got like E equals HF type stuff.
When there's frequencies, you should suspect the complex numbers are there.
It's useful in electrical engineering.
Why?
Because you're dealing with a bunch of waves.
You've got a bunch of cyclical processes and frequencies and so it's natural to model them
with these kinds of numbers. So the engineers adopted the complex plane,
the complex numbers after the fact, right?
They didn't say, gee, we need a way to do this.
Let's invent it.
They said, we don't know how to do this.
And a mathematician comes up, here's a way.
I mean, of course, I'm exaggerating that.
I mean, do you want to know where complex numbers came
from originally?
Because I think it's sort of a lie that we tell in schools where we say, you know, there's
no such thing as a square root of negative, but like pretend like there is and like mathematicians
just love pretending things.
It actually cuts to something we were talking about earlier, which is when people were solving
cubic equations and they wrote down a cubic formula that thank God neither of us had to
memorize.
I don't thank God for that.
I'm very picky about what I thank God for.
He's like, thank math.
Thank the education system.
But if you would ask someone,
hey, solve the equation x squared plus one equals zero,
they would have said, there is no solution.
Like obviously there's no,
we're not going to make up a solution,
that won't do anything.
But there were certain cubic equations
where when you tried to use the formula, you had
a real number answer.
So real numbers in, real numbers out, never any whiff of the square roots of negatives.
But when you use the formula, you can find that real valued answer if you take seriously
the idea that somewhere inside that formula there's a square root of a negative and it
all cancels out at some point.
But while you're working it out, you're engaging with these square roots of negatives.
So for a while, for mathematicians, they're like, oh, it's this one weird trick that kind
of works.
I don't know what it means, but it seems to work for solving cubics.
And I think one of the reasons they were called imaginary is because it took a long time for
people to take them seriously.
They just thought it was this notational trick.
And the word imaginary was kind of derogatory. It wasn't like, hey, we want to teach kids
this, what should we call it?
It was like imaginary girlfriend.
But then it took much longer for people to realize how they're useful and the idea that
they have these cyclic properties that make them really useful for any kind of math or
physics that involves waves and cycles. And so, that's what we're stuck with.
Hello, Dr. Tyson, Lord Nice, Mr. Brown.
This is Brandon from New Jersey.
Is there any relation between the conformal geometry and the Pythagorean theorem?
Recently I've learned about some circle inversions and it seems to me that these inversions are leveraging the
Pythagorethum theorem to maintain the symmetry of points across a line after
curving it. Is this at work in conformal geometry as well? I'm completely
confused but wow what do you get out of that? Look at that man. Let me just answer as well? You and me at a blackboard. The lesson will be better for everyone if we don't engage with what you want
to engage with right now.
If you wanted, we can try describing
what conformal geometry and circle inversion are.
All right, can you do it in 20 seconds as a challenge?
And if we understand it, who cares?
If not, then now we have something to go look up,
which is even more fun.
Go ahead.
All right, go.
When you look at yourself in the mirror,
you see a reflection of yourself.
It's like a different version of yourself.
Sometimes you can use that to solve problems like solving your hair, and things like that.
In mathematics, there's a different kind of mirror that sometimes they use where they
pretend like a circle is a mirror and reflects everything from the inside to the outside
and the outside to the inside.
It's the special transformation of space.
And there's certain geometry problems where they look hard but then when you reflect it through a circle like this which is a
really weird mind-warping motion it turns the hard problem into an easy
problem that's called circle inversion some of the aspects about that I can't
describe conformal geometry but that's what circle inversion that's the vibe of
it and we won't describe specific problems there, but if you're curious on the vibe,
it's treating a circle as a mirror.
Cool.
That was fascinating.
I mean, I don't,
I get with the circled inversion what you're talking about.
I don't know what it's used for and I don't understand.
Yeah, I got to give actual examples
for that to carry teeth, but you know, a little too long.
He was still showing off though.
Yeah, he's showing off without a doubt.
Brandon was showing off, okay?
But guess what?
I'm glad, I'm glad, you know, I've never even heard of conformal geometry until just this
moment so I'm happy that Brandon was showing off because now it's fodder for look up, which
is great.
Ethan Stepp, and Ethan says, hello, Dr. Tyson and Mr. Grant.
And Lord Nice, if you're there, hello. I love these people, man.
Freaking love these people.
He says, my name is Ethan from North Carolina.
I love math, but sometimes I wonder
how these things get figured out.
My question is, how on earth did someone come up
with tensor products?
Okay, I think the easiest way to describe it,
it doesn't quite capture what it's about,
but like a vector, we have a list of numbers
that you might write as like a column of numbers.
A matrix, you've got this two dimensional grid of numbers
used in computer science all the time,
it's how machine learning works,
but sometimes you want a three dimensional grid of numbers,
just as the way to hold your data.
Imagine like a three dimensional grid, each cell in that as the way to hold your data. Imagine a three-dimensional grid.
Each cell in that grid has a number.
You might call that a tensor.
That's the computer scientist way to answer what a tensor is.
But then in physics, the use of certain objects which can be represented with tensors like
this is relevant for general relativity and describing the curvature of space. There's also other corners of math where you have objects that could be described, like you could give them a coordinate
system where you want to have like a three-dimensional grid of numbers like this.
And to answer the question like where did this come from or who would come up with it,
I think usually it's once there's a very specific problem that you're dealing with where you realize
the data that represents this is something that naturally organizes itself into a three-dimensional
grid or a higher dimensional grid of numbers, it's just a natural way to try to even hold
that idea in your head.
But somebody's got to be clever enough to see that need and then come up with it.
Not everyone is that clever.
Yeah.
There would be a prisoner of the known math of the day.
And you need someone who can step out of that and say, I have a new way to think about this
problem.
I mean, some physicists will joke that one of Einstein's greatest contributions to physics
was his notation for tensors.
And like he had a really nice notation for how to even write it down, which lets you
think about it clearly on a blackboard and such, which is a joke, of course,
but it does cut to the fact that they're a central object
that it takes a clever mind to even like represent
in a useful and manipulable way.
This is not T-I-J, is that?
Yeah, this is when you're like the Einstein summation
notation where you wanna, yeah.
Yeah, it's a simplified thing.
It makes the equation look way simpler
and more tame than what's actually going on.
Well, I'm going to tell you,
you guys just made it look way simpler to me
because I have no idea what you're talking about.
Okay?
I'm just going to be honest.
I'm sitting here, and it's rare that I am lost
when in a conversation, but this one is like out there, man, which is very cool.
Plus he's got seven million followers on his YouTube,
so he's doing something right.
He's doing something right, yeah.
Hi Grant, hello Dr. Tyson.
I'm Akia, and I'm originally from India,
but I live in San Francisco.
I heard the last podcast when Grant was on,
and I was on my way to Death Valley
for a dark sky festival for stargazing night
And it was a delightful experience
We stopped along the way at a cafe and had a lovely cup of tea
I had George healing my friend Mary had no no I'm making this part
Including all of your information about your trip my question to you grant is
Why is there no new
revolutionary paradigm in mathematics
like calculus and algebra that is uncovered today
as opposed to the last millennium
when many fewer people were working on mathematical problems?
Thank you so much for both of you
and for popularizing science.
It sounds like a diss.
Where is the next branch of math?
Right.
What's up with that?
You guys have done nothing.
I mean there's definitely been a ton of development in math and new fields developed.
Most of the math that exists.
So okay, algebraic geometry, there's this very interesting phenomenon that's been happening
in the last century or so, but even the latter half of that, if someone tried to think of
like a grand unified theory of math, something where you're trying to understand what are
these weird connections that come up in seemingly very disparate parts, one of those fields
that tends to take that perspective coming to stepping back and saying, hey, what if
prime numbers and like functions were really living in the same kind of world,
and the facts that we know about one tell us facts about another? A lot of the people doing
algebraic geometry, that kind of fits in there. You also have a pretty big revolution in the way
that people think about math, where there is a thing that's called category theory, which is extremely hard to explain,
and it's kind of like a language,
it's sort of like a new language
with which mathematicians think about their work
that didn't exist 100 years ago.
It's very much a different way of thinking,
and it's just quietly happening among these circles,
not in a way that's very popularized,
it's never gonna show up in your high school calculus class,
but it's absolutely like a new thing.
Okay, but you're saying it exists within math,
but none that have shown up in our K through 12 textbook.
Nor should they.
Like, I don't think you should shove category theory
into a K through 12 textbook.
I got to say, what Grant's doing right now is
he's throwing shade at all those mathematicians back there
that Akita is talking about.
He's like, our stuff is so complex right now
that yeah, you can't even learn it, okay?
No.
No.
No, okay.
Only three people can learn it in the whole world.
That's it, baby.
No, I know you're joking, I know you're joking,
but I hate when this is kind of how things come across
as like, oh, this is so complex.
It's more like, so you have certain people
doing a certain job, like a research mathematician
trying to find proofs.
This is tools for that job.
It's not a great tool for other jobs,
like maybe writing programs and using mathematical modeling
for the simulation that you're running.
Not as good a tool for that job.
But for their job of writing proofs,
it's like, here's this new tool.
It's come into invention.
Once you want to pursue that job,
we can make it as approachable as we want.
But because I don't think everyone should do that job,
like we shouldn't put it into K12.
And there's a ton of stuff that's too complex to describe
that's vocational, right?
If you want to understand like the exact way
that injection molding works or something like that.
It might be something that's inappropriate for a podcast,
not because it's like highfalutin, you know,
super math brain.
It's just because, hey,
things that are very peculiar
to one job tend to involve a lot of assumed jargon
and a lot of assumed context from the people learning it.
And it's just not meant to be.
Akiya, guess what?
It's there, but it's not necessary for,
you want a need to know basis.
You want a need to know basis, Akiya.
And that's the, there we go.
All right, this is Gina.
She says, hello smarty pants.
I was,
okay, I love it.
She says, Gina Martin says,
I was recently learning about circles and pi.
I learned that the circumference of a circle
can never be a whole rational number.
I am having such a hard time wrapping my head around this,
pun intended, wrapping my head around, okay.
Could you please explain this a little better for me?
Thanks, Gina from North Carolina.
Wait, wait, wait, wait, is she correct?
Can't you have the diameter be an irrational number and end up with the
circumference rational?
Yeah. So, to be clear, if you want the circumference, the circumference can be anything you want,
it could be the number five. I think the intention of the phrasing was that the ratio of the
circumference to the diameter could never be, or like if the diameter is a whole number,
the circumference will never be rational.
However, you want to phrase it, it's the relationship between those two that's fundamentally irrational.
So it's like a game of whack-a-mole.
You make one of them a nice number, the other one looks ugly.
You make one of the other one nice, the first one becomes ugly.
Oh, ugly.
It becomes irrational.
It becomes hard to write down.
And so it's a very deep question to try to say why is pi
this ratio between a circle circumference and its diameter? Why is
that an irrational number? There is not a podcastable answer that I can give. This
might not be satisfying but instead of talking about circles let's talk about
squares where if you have a square and it's got a side length of 1 and you ask
how far is it to get from one side length from one corner to the opposite corner.
That ends up being the square root of two.
This is something that follows from the Pythagorean theorem.
This is another situation where this geometric length has an irrational relationship with
the first length we drew.
The ratio between that diagonal and the square side length is the square root of two.
Now that is irrational.
It's also much, much easier to prove to you
why it must be irrational.
And if you will indulge me,
I think it's possible to do this in like 45 seconds
and we can see how this goes.
All right, you ready?
Here we go, prove the square root of two is irrational.
Assume that you could write it down as a rational number.
Like maybe you think, oh, maybe square root of two is going to be,
I don't know, like five divided by three,
or maybe something more complicated,
like 153 divided by 311.
Surely, I can find big enough numbers
that'll make this work.
And I say, whatever you choose,
we'll write it down as P over Q.
We say that's the same thing as the square root of two.
What that would mean,
first of all, let's assume that it's fully reduced.
So if you wrote something down like four divided by two,
you could reduce that to be two divided by one.
So there's no common factor,
you can reduce this thing down.
So if that was true,
P divided by Q is the same as the square root of two.
By definition, you're saying that P squared
divided by Q squared is equal to two. So that's what it would, by definition you're saying that p squared divided by q squared is equal to two.
So that's what it would mean by definition.
So that means when you multiply everything out by the bottom, that q squared, p squared
is going to equal two times q squared.
So if you could come up with some numbers where it was true, you must admit that p squared
is the same as two times q squared.
That means that p is an even number because it's two times something. p
must be an even number. Yeah. And so you're like, okay, I don't know. Let's call p, you
know, two times k or something, right? It's some even number. If you then write this down
algebraically and you replace it with two times k, you're going to conclude that q also
has to be an even number because when you take that key equation P squared equals two times Q squared that
Ends up looking like four times K squared equals two times Q squared you divide some stuff out and you say hey Q
Also has to be an even number
So you must conclude that P is even you must conclude that Q is even
But we assumed at the start that it was a reduced fraction both of those numbers couldn't be even otherwise it wouldn't have been reduced
So there there cannot be a way to write it as a fraction, because otherwise you end up in
this infinite regress where somehow both of them have to be even, but if you reduce it
down now both of those have to be even and you'll never get to a coherent answer.
It's just a little weird that to get this irrational number you have to take the ratio
of two numbers. That's a weird fact.
So, this is a common mathematician tool. They say, oh, you want to take the ratio of two numbers.
This is a common mathematician tool.
They're like, man, this problem's really hard.
They have a big ego, and so they want to say,
hey, it's not that I can, what you do is you say, I'm gonna start by assuming it's possible,
like writing some notation to say,
what if it was possible?
What would follow from that?
And then you come to some kind of contradiction.
You say, so see, if we assumed it was possible,
we land on this thing that could never be,
therefore our assumption was false.
So that's a very common mathematician thing.
I love that, what you just did.
I love starting with the square.
That's very cool. I'm Tien from Vietnam.
If your to journey to Flatland, what shape would you use?
I mean, would you choose to be and what activity would you like to do there?
Thank you, long time fan of StarTalk and both of you.
Tell us about Flatland.
I happen to have a copy right here on my desk.
Oh, really?
That's great.
Right there.
Yeah, I've got one over on the shelf over here.
So, this is a very classic book where the author, had you imagine a world that's just
two-dimensional?
So here in three dimensions, you can look left, right, up, down, in and out.
But he said, what if you were just on this two-dimensional world and that's all the world
was?
And so you have a bunch of creatures there.
And he was making this analogy to say, like, wouldn't it be really hard to describe three-dimensional
shapes to them?
Like, if you have someone who lives in flatland and you want to describe what a cube is or
a sphere or like a donut,
there are these shapes you just really can't describe it to them. And then the purpose of
the book was then to say, hey, if there's geometric shapes in four dimensions, it is as hard to
describe to us as it is for us to describe to flat landers. But to the question on what shape
would I be in Flatland.
I mean, it's kind of basic maybe, but circle seems useful. You can roll around, everything's nice and symmetric.
You can do circle inversion, which the other patron boys
are only going to appreciate, but that's probably all I got.
As I remember the story, the more sides you had,
the more aristocratic you were. Ooh, look at you. So a triangle would be like the story, the more sides you had, the more aristocratic you were.
Ooh, look at you.
Right, so a triangle would be like the lowest,
the scum of the earth.
Mm, those triangles, I can't believe that.
Trying to move in here, they know better.
And then the squares, and then pentagons, hexagons.
So I think I'd be a hexagon.
Okay.
I'd be a hexagon.
You want to tessellate the plane?
Yes, I want to. Ah, look at that. Yeah, although any shape can tessellate, right? I'd be hexagon.
Although any shape can tessellate, right?
Well, not any shape.
If you look at Escher paintings,
isn't that tessellation?
Two shapes intersect, so the difference is,
I'm what they call a regular polygon,
where all my sides are equal to each other.
Then it's only the hexagon.
But I think tessellation is all shapes
that can do that is called tessellation, isn't that right?
Well, not any shape can tessellate.
So you're right, the pressure shows
you've got this infinite family that can tessellate.
I agree, but you have like angels and devils tessellating.
Yeah, yeah, yeah, yeah.
But that's called tessellation, isn't it?
Mm-hmm, yeah, 100%.
Right, right, so now if you're going to restrict yourself
to a shape, to a polygon, what's called a regular polygon,
then we're limited, right.
But so I want to be a hexagon
because you can tile a floor with a hexagon.
Yes, you can, yeah.
And get other people to tile with you
and you can snuggle and everything fits all.
Fits very, very snugly together.
Snugly.
Very cool.
So what shape would you be?
Now that we're talking about tiling,
there was a whole tile that was discovered
a couple years ago that's a single tile
that tessellates in a non-periodic way
and it can't tessellate in a periodic
but it only tessellates non-periodically.
I heard about that.
But is that a regular polygon or is it just some other shape? No, no, no, no, it's a wild, but it only tessellates non-periodically. I heard about that.
But is that a regular polygon or is it just some other shape?
No, no, no, no.
It's a wild...
It's not that weird a shape.
It looks like a hat, kind of, people call it.
But what's cool is it was discovered by an amateur.
So people didn't know if there was a shape that tiles things non-periodically or that
tiles things non-periodically without having any periodic tiling is the
like technical question.
But it's just like an interesting tiling question.
An amateur found it and it became a little fun celebrity of the math internet for a couple
months back then.
So what's the difference between a periodic and a non-periodic pattern?
Great.
So most of the patterns you can think of are periodic.
It's kind of what we mean by pattern.
Almost yeah.
Like if you shift the whole picture and it looks identical
So you take your hexagon tiling and then you like shift your view over by one hexagon. It looks identical
So that's what we would mean by periodic
It wasn't even known that you could have a non periodic tiling for a while
But Penrose who very famous for physics reasons
He found a way to use these two tiles that each look like a rhombus to have a pattern
that fills all of space, but it never repeats.
So it's a describable pattern.
You can describe what it should be, but it never repeats.
So when you shift your viewing point, it will never look identical.
So there's no way to shift it to be the same as what it once was.
Yeah.
It's kind of like how the digits of pi are these irrational numbers.
They don't repeat themselves.
They just go on and on in a predictable way, but not that repeats
itself. It's the geometric equivalent of that.
Okay. Very cool. This is William Walker. And William Walker says, hello gentlemen, from
the Florida Panhandle. I've heard it said that mathematically we know properties, some or all, I'm not sure, of dimensions higher
than what we observe. Could you please elaborate upon this? What can we say about these dimensions?
Yeah, how do you get there?
Great, great, great, great. I think this is one of the big misconceptions when mathematicians
talk about higher dimensions. People assume that they are talking about something that should be physically realized.
So ultimately, when you're doing math,
sometimes you might have something that can be described
by multiple numbers.
You have some system like a little particle moving around,
and you describe its velocity with some list of numbers
and its position with some list of numbers.
And you often find it useful to take all your numbers
and just list them together. And if you have a list of three numbers,
you could think of it as a point
in a three-dimensional space.
If you have a list of two numbers,
you could think of it as a point
in a two-dimensional space.
Uniquely.
But the math-
Uniquely, yeah.
And mathematicians and physicists realize like,
hey, sometimes we're solving a problem
and we have a list of like four numbers or five numbers.
It was really useful to be able to like visualize what was going on when it was a list of three numbers numbers or five numbers. It was really useful to be able to like visualize
what was going on when it was a list of three numbers
by having this unique association between a triplet
of numbers and a point in a 3D space.
They're like, why can't we do that?
Why can't we say there's some abstract
four dimensional space, not in like physical reality,
but that's just gonna represent whatever problem
I'm solving where there's a quadruplet of numbers
that come up or in machine learning these days, days, when a large language model reads your text, the
first thing it does is it turns a given word into a really big list of numbers, like tens
of thousands of numbers.
And it's very common for researchers to think of that as a point in an insanely high dimensional
space and to use geometric ideas to describe what's happening to it through the model.
But of course we're not saying
there's like a 12,000 dimensional space
in physical reality, it's just that it's a nice way
to describe lists of numbers.
I have on my shelf here.
So these dimensions are placeholders.
I got on my shelf here a Klein bottle opener.
I love it.
So this is an attempt to represent
a four dimensional object in three dimensions.
Yeah.
Yeah, so Klein bottles are something that,
they're most comfortable in four dimensions.
This is where they want to live.
And if you try to make them live in three dimensions,
they have to like, unnaturally cross through themselves.
There's no way to put it in three dimensions
without it crossing through itself. So the Klein bottle is a bottle that has
no inside. Yeah, I think that's a fair way to say it. There's no, you can't distinguish the inside
and the outside. Yeah. I had some friends in college who got in trouble for trespassing in
a certain building and they're like, maybe as part of our defense, we go to the fence outside the
building and we apply one twist to it so that the whole fence is a Mobius strip
and this is another one of those shapes
where there's no clear notion of an inside or an outside.
Like then we can argue to the authorities
that we couldn't have been inside the trespassing area
because there's no coherent notion
of the inside of the relevant area.
Were these bored Stanford students?
Yeah, these were.
I don't know how bored they were,
but they were creative Stanford students.
Not creative enough not to go to jail.
Because you're still going down.
But that's clever.
You'd have to flip the fence, but then reattach it.
Reattach it.
Yeah.
Yeah, exactly.
So, you know, while you're doing whatever you're doing, trespassing there, just make
sure that as you leave, you cut the fence, you twist the fence, you reattach it so that it's a Mobius fence and then your
defense is solid.
So a Klein bottle is a four-dimensional version of a Mobius strip.
Kind of, yeah.
I don't love that description.
I mean, if you try to take a Mobius strip and you take another Mobius strip and you
try to glue their edges together, you'll get a Klein bottle.
It's a very mind-warping thing to try to think about.
It's analogous to a Mobius strip in that they both are non-orientable, meaning you have
this notion of no clear inside or outside, but they're different.
A Klein bottle is a closed shape.
It doesn't have an edge.
Mobius strip has an edge.
So topologically, they're pretty different animals, but they swim in the same waters.
Well, it doesn't have a 1D edge, but this has a 2D edge,
which would be a surface.
A surface is an edge in four dimensions, isn't it?
In the same way, the 1D edge of a Mobius strip
is an edge in three dimensions.
So if you live on the Earth, right, and you try to walk
to find the edge of the Earth, it's a sphere.
There is no edge.
You're never going to go to the edge
where all the water's falling off, right? If it was a flat disk, you could walk to where the edge of the Earth, it's a sphere. There is no edge. You're never going to go to the edge where all the water's falling off.
If it was a flat disk, you could walk to where the edge is.
If you're a little ant and you live on a Mobius strip,
you can walk to the edge and peer off the edge at some point.
If you're walking around the Klein bottle,
you never hit an edge in that way.
And mathematically, we call it a closed surface in this way.
So they're both surfaces. They're both 2d
Okay, so there's an important distinction closed surface. That's the key time for just one more question. Okay. All right
Let's go to our old friend Kevin the sommelier. Oh, okay, and Kevin says his last name the sommelier
Middle name is the last name of sommelier
Hey, he says this hey Neil has touched the three-body problem in an explainer episode,
but would there be another branch of mathematics
that hasn't been discovered yet that could solve it
just like Newton did with motion
in honor of Sir Isaac Newton?
There is a Spanish wine called Principa Mathematica,
which is a bright white wine made with Ciariello.
I'm going to find that wine.
Yeah.
So Principia is his greatest work.
It's the mathematical principles of natural philosophy.
Uh-huh.
Sensibly abbreviated Principia.
Principia.
Yeah, but Principia Mathematica,
if you want to give the full thing.
Oh, okay.
This is his greatest work.
If there's a wine with that name, I'm going to find it.
Thank you.
There is a wine with that name.
Sommelier.
So I like this question because at what point do you say it's unsolvable and at what point
do you say the person who's brilliant enough to solve it is yet to be born and to apply
their genius to it?
Yeah, I would say there's two different ways
to think about if a problem's hard to answer.
So in the case of Newton modeling the planets,
it wasn't even known what the right math to put to it was,
what mathematic model you would use
to try to make predictions.
And his big contribution was to invent
the appropriate field of math
that you could use to then make predictions.
A branch of mathematics that you're not inventing today.
Just thought I'd rub that in again.
Okay.
Right.
I mean, the three-body problem feels so different because it's not that it's like we don't know
what math should describe it.
It's instead saying, we know, it's an intrinsically mathematical question.
You say, given this piece of math that's describing it exactly and it's Newtonian calculus, it's
known that you cannot predict what's going to happen if you have a little bit of error
in your initial predictions.
This was the big surprise of chaos theory where initially you might think, hey, if I
know how to solve an equation or if I have some equation, I'm sufficiently smart about
it, then if I know the initial state,
and I just see how the world evolves
according to that equation, I can predict the future.
And then chaos theory said there are these situations,
including the three-body problem,
where even if you exactly know what all the solutions are,
if you have a little bit of error in your measurement,
that error blows up so quickly that subject to that error, the possible states
you could end up with after a pretty short amount of time, spans such a wide space of
possibilities that effectively the outcome is unpredictable.
So unless you had infinite precision, which is just that's not how science or engineering
works at all.
Or planets move.
So the result is telling, it's not that it's unknown,
like what the answers will be, it's known that the answers
are unknowable in a certain way, right?
It's known that it'll be chaotic in the sense
that final outcomes are very sensitive.
So that's not the type of problem lending itself
to this issue.
Right. Yeah.
We just need a smarter person to come along.
No, because the idea is this.
The answer is, it is unknowable.
That is the actual solution.
Oh, I got you, that's the answer.
That's the answer.
It's not that we can't solve it.
We did solve it.
We did solve it.
Exactly.
And the solution is, this is unknowable.
You agree with that, that's brilliant.
Yeah, that's a great summary.
I think that's a great summary of chaos.
I've earned my keep here.
I can go home now.
Just barely.
I know what time is it.
Really, Chuck, on the last question?
One of the things I like most about mathematics
is you get to peer through doorways
in advance of actually stepping there.
Because the math, as a model of reality, speaking as a scientist,
allows you to explore the world without ever leaving your chair.
If the math is a proper model of the physical universe,
then you have the power of a God in your hand by making predictions.
If they come true, that gives you that much more confidence that the math and the universe are one.
And with that power, I thrive on thinking about higher dimensions, a topic we talked about.
What do things look like in four dimensions, five dimensions, six dimensions?
You can't picture that in your head.
No.
Our brains evolved on the planes of the Serengeti trying to not get eaten by lions.
We don't have the capacity to think that way, but we have the mind power to calculate that way and
to give us all the information we know about higher dimensions and other places
we have yet to visit. And that is a cosmic perspective. So again, tell us your
YouTube channel. Yeah, so the YouTube's channel, it's named 3Blue1Brown,
an admittedly weird name. You
could also search 3B1B and a lot of topics that we've discussed here, flavors of them
show up on that channel.
Excellent. And how else do we find you on social media?
3Blue1Brown on whatever your favorite social app is these days.
Okay, now the three and the one are numerals, and the blue and the brown are words, correct?
Very hard to describe.
The numerals mixed with the numbers, yeah.
Or just 3B1B, if you search that,
you can usually land on it.
Okay, well just congratulations for your success,
and as I say, doing God's work.
Yeah.
Because God can't divide by zero.
Nope.
So, we hope to see more of you.
Keep up the work, and we need more math fluency in this world.
Yes, we do.
God, please, just infect the whole country.
Can you please?
Do my best.
All right, Chuck, always good to have you, man.
Always a pleasure.
All right, this has been StarTalk Cosmic Queries, mathematics edition.
Neil deGrasse Tyson, keep looking out.