Episode 92 - Kate Stange

My Favorite Theorem

Evelyn Lamb: Hello and welcome to my favorite theorem, the math podcast with no quiz at the end. I'm Evelyn Lamb, a freelance writer in Salt Lake City, Utah, where it is gorgeous spring weather, perfect weather to be sitting in my basement talking to people on Zoom. This is your other host.
Kevin Knudson: I’m Kevin Knudson, professor of mathematics at the University of Florida. I don't know, Evelyn, I saw the pictures on Instagram over the weekend and it looked cold in Utah. You wrote that you rode a century, right?
EL: Metric century.
KK: Okay. Metric.
EL: Just in case — you know, I don’t want people to think I'm quite that hardcore. Yeah, at least at this point in the season. Yeah, I hadn't managed to ride more than about 25 miles since last fall because weather, travel, just things conspiring against me. The week before I was like, I really need to get 30 or 40 miles in on Monday. And then it was, like, 20 mile an hour winds and sleet and I was like, well, I guess I'm just going into this cold, but it was fine. It was actually gorgeous weather. It was a little chilly at the start, but better than being too hot.
KK: Well, you know, the muscle memory takes over, right? So you can do — I mean, 62 miles isn't that much more than 25, really, once you have the legs, so congratulations.
EL: Yes, thank you. Well, we are delighted today to be joined by Kate Stange, who is in the Mountain Time Zone, something that I always feel thrilled about because I'm constantly converting time zones when I'm talking to people, and finally, someone I didn't have to do that for! So Kate, other than being in the Mountain Time Zone, what would you like to tell us about yourself?
Kate Stange: Oh, geez. Um, well, I'm also a cyclist. And so I'm jealous hearing about your rides.
EL: Wonderful!
KS: Here in Colorado we have we have this ride called the The Buff Classic. And so it has a 100 mile option where they close Boulder Canyon so that you can bike up the canyon without any cars.
EL: Oh, wow.
KS: Then you bike along the peak to peak highway. It's just wonderful.
EL: Yeah.
KK: Yeah. That sounds great.
EL: So you're at CU Boulder?
KS: Yes. And it's run from the campus. It starts right outside the math department.
EL: Oh, perfect. Yeah, just drop your stuff in your office and hop on and ride it?
KS: Yeah.
EL: Yeah, great. Well, we are thrilled to have you here today. And I guess we can just dive right in. What you're on what do you like to tell us about?
KS: My favorite theorem, at least for today, is the bijection between quadratic forms and ideal classes.
KK: That’s a lot of words.
EL: Yeah, and I'm so excited to hear about this, because I am honestly a little nervous about both quadratic forms and ideal classes, and a little embarrassed about being nervous about quadratic forms, not so much with ideal classes. So yeah, can you tell us a little bit about what that all means?
KS: Yeah, sure. So quadratic forms is probably what sort of comes first in the story, at least sort of the way that the mathematics tells it, and also probably the historical way. And so a quadratic form is just a polynomial with an x2, a y2 and an xy. So it's like 3x2 + 7xy − y2. So that's quadratic form. And, as number theorists, one of the things that we're most interested in studying is what are the integer solutions to polynomial equations? And so first you start with linear equations. And there's a wonderful story to do with Euclidean algorithm and stuff there. And then you move on to quadratic. And really, these are sort of some of the first equations that you would start studying next, I guess. And so they go back to the classical days of number theory, Gauss and Euler and everybody. And, yeah, so they come in, what happens is that they come together in families. So different quadratic forms, you can actually just do a change of variables. And it'll look different, but it won't really be too different, particularly if you're interested in what numbers it can represent when you put in integers. So say I take x2 + y2, which is the simplest one, if I put in various different integers to that I get various different integers out. And then if I do a change of variables on that, just a little change of variables — like maybe I change x to x + y, but I leave y alone — the formula will look different after I do that change of variables, but as I put in all integers and look at all the stuff I get out, those two sets, the in and out, they're going to look the same. And so we kind of want to mod out, we want to ignore that difference. So I'm really thinking of equivalence classes of quadratic forms. So that's the first object.
EL: And that change of variables is kind of the only equivalence class thing that happens with them?
KS: Yeah. Yeah. Because they could really behave differently between the different classes.
KK: And you only allow a linear change of variables, right?
KS: Yes, exactly. Yes. Thank you.
EL: Yeah. Okay. So now, ideal classes.
KS: Now ideal classes. So this is an interesting one, to describe where it comes from, I think. So there's sort of — if you think about the history of math, I would say there are sort of two versions, there are sort of two histories of math at the same time. There's one, which is sort of the human history, which is fascinating and human and quirky. And then there's sort of the way that the math would like to unfold to human understanding the way that as any human coming to it, they might discover the pieces of the mathematics. And I don't know too much about the details of the human history. But in terms of how you might discover this, if you're just looking at the integers, you are interested in how they behave, you discover things like prime numbers, you've got addition, you've got multiplication, you've got powers, you might ask how these things interact. And at some point, maybe when complex numbers are discovered, you think about whether there are possibly other collections, or other number systems, other collections of numbers in which you could do the same kind of thing. And so one of the first examples of this would be what's called the Gaussian integers, which is where you take complex numbers, I don't know whether I should dive into complex numbers, but you take complex numbers with integer coefficients.
EL: So that means things like 1 + 2i or something. So the i and the 1 both have integers in front of them.
KK: Right.
KS: Yeah, exactly. And so this is a collection of integers, kind of, right? And we ask things like, okay, are there prime numbers? And so it turns out that there are in that system, there are Gaussian prime, so, like, 1 + i is a prime number. And so you kind of start to develop this whole theory that you have for the integers. But what you find is that in some of these systems, you lose unique factorization. So we love unique factorization in the integers, right? Every integer, up to reordering the factors and maybe putting on a minus sign, you have always a unique factorization into prime numbers. And in the Gaussian integers, that's true. But in some of these other systems, you lose that. And so what people tried to do is to try to fix it. And it turns out, the way to fix it is to add in what were I think, originally called ideal numbers. They were thought of as numbers that should be in the system that weren't in the system. And what they actually were were collections of numbers. They were sets of numbers instead of individual numbers. And the idea here is that, say, you were to take — like in the integers, if you took the number two, you could replace that idea, that idea of two-ness with just the collection of even integers. And so that's an ideal now. Instead of a number, it’s an ideal, and it's really carrying the same information. But now it's a subset. And so by moving from individual elements of the ring, of the collection of numbers, you move to subsets of them. Now you have more things, and so now you can recover unique factorization in that world. So those are ideals.
EL: Yeah. And so the Gaussian integers do have unique factorization.
KS: They do. Yeah.
EL: So this — actually, I kind of forgot, but recently, this came up in something I was writing where I wanted the example to be the Gaussian integers so bad because it doesn't have any square roots in it. But then it didn't work because it isn't true for that. I was trying to show how unique factorization could fail, but I didn't want to have to use square roots. But as far as I know, you can't do that. So then I fixed it by putting a square root of negative five in there and hoping that people would be patient with me about it. But yeah.
KK: So that's the example of one where you don't get unique factorization, right? So you take the integers and you join the square root of minus five. That's one example.
KS: That’s one of them. Yeah.
EL: And then it's like two and three are no longer primes.
KS: So if you multiply (1+ √ −5) × (1− √ −5)
KK: You get six. Yeah.
KS: You get six, which is also two times three. And those are two different prime factorizations of six.
KK: Right.
EL: Yeah, but it's so fun that you can do that, and I like your way of putting it where regardless of how these ideas actually formed, you can as a human, looking at some of these basic pieces like primes and then or primes and integers and square roots and things, you can kind of come up with this, like, what happens if I do this? And create this new thing where this this property that I know I always assumed — like unique factorization, when you're growing up, you know, when you take math classes in school and stuff, it just seems like so basic, like, how could you even prove that there's unique factorization? Because how else could you factor anything?
KS: Yeah, exactly.
EL: It feels so basic.
KS: Yeah. And this is what happened, I think, historically, too, is that people didn't expect it to fail. And so they were running into problems and it took a while to figure out that that's what was going wrong.
KK: Wasn’t this part of, was it Kummer who had a reported proof of Fermat's Last Theorem, and he just assumed unique factorization?
KS: That’s what I've heard, although I never trust my knowledge of history. Yeah.
KK: It’s probably true.
EL: Well, and there are a lot of good stories. And they may or may not be true sometimes. But yeah, okay. So we've got these, these two things.
KK: Yep.
EL: The quadratic forms and the ideal classes. So yeah, I guess either historically or mathematically, what is this connection? And how do you know that these two things are going to be related?
KS: Yeah, so they seem like different things. So I think quadratic forms were studied earlier. And at some point, people noticed that quadratic forms had an interesting property, which is that sometimes you could multiply them together and get another quadratic form, which is kind of hard to explain. But like, if you actually wrote out (x2 + y2) × (z2 + w2) and you multiplied that all out, you'd have a big jumble. But then you could factor it out. So it looked like, again, a square with some stuff inside, z’s and w's and whatever inside the brackets, plus a square. And so this meant that sometimes if you picked your forms correctly, and they had this sort of relationship, then if you looked at the values they represented, the numbers that can come out, when you're putting integers in, you would take that set of things the first one represents and the set of things the second one represents, and then you’d look at what the third one represents, and it would represent all of the products of those things. So there was this definite relationship, but the way I'm describing it to now is a little awkward, because it's a lot of algebra. But this is, I think, what was noticed first, somehow. And again, I might be mixing the human story with how math tends to want to unfold. I don't know exactly the history. But anyway, so you notice that there's this relationship. And that's kind of reminiscent of an operation, like a multiplication law. And what happens is that, in fact, that's coming from the fact that these ideal classes, each one of them — sorry, my mistake — so it's from the fact that each of these equivalence classes of binary quadratic forms, each one of them is associated to an ideal. And the ideals as the sort of generalization of the idea of number, they can be multiplied together to get new ones. And so on the ideal side, it makes sense that there's an operation because you're already living in a number ring where you've gotten an operation. But on the quadratic forms side, it's a surprise. And so that's one of things I like about this theorem is that you see some structure and you want to understand why. And the reason to understand why is just to change your perspective and realize these objects can be viewed as a different kind of object where that behavior is completely natural. Yeah, so that's one thing that I like about it.
EL: And does this theorem have a name or an attribution that you know?
KS: Oh, it's such a classical theorem that no, I don't know.
KK: Right. It's just the air you breathe, right? So what's the actual explicit bijection? So you've taken a quadratic form. What's the corresponding ideal?
KS: Well, actually, the other way is a little bit easier to figure it out.
KK: Yeah, let's go that way.
KS: So let's take the Gaussian integers, okay. And in the Gaussian integers, you've got — for your ideal, so think of it as a subset of the Gaussian integers. But because it's an ideal, it has the property that it has the same shape as the Gaussian integers. I actually usually like to draw a picture. So I'm going to try to draw a picture just out loud. So if you think of the Gaussian integers in the complex plane, they fill out a grid, right? It's all the integer coordinates in that plane. So that's a grid. And if you want to see what the ideals are, they’re subsets that are square grids as well, but fit inside that grid that we started with, maybe rotated or scaled out.
EL: Okay.
KS: But they're square again.
EL: Okay.
KS: And so, what you can do is with this example, specifically, you can take the norm of each of these elements in the Gaussian integers. So the norm of a complex number, usually I think of it as the length from the origin. But I don't want to do the square root part. So if I have a Gaussian integer x + iy, I'm going to take x2 + y2, and that's the norm.
KK: Okay.
KS: All right. And so if I take the whole Gaussian integers, which is itself an ideal, that's one of the subsets that is valid as an ideal, then if I take all of the values, all the norms of all those elements, that's all the values of x2 + y2. So from my collection of integers, I take all of the values and that's actually a quadratic form.
KK: Okay.
KS: Okay?
KK: Okay.
KS: And so you can do this with the other ideals as well. So for each one, you look at the norms of all of its elements, and that is a quadratic form and the values of that quadratic form?
KK: Right. So the Gaussian integers are Euclidean, right? So it's PID, right?
KS: It is. It’s a principal ideal domain.
KK: So everything's generated by one element, basically every ideal?
KS: That’s right.
KK: So that makes your life a little simpler, I suppose.
KS: Yeah. So the ideals, in that case, really, they're not so different than the numbers themselves. This is one of those ones where you don't have to go to ideals. But by going to it, you think about instead of just, say, 1+i the number, you think about all the multiples of 1+i and you take all of those, and you take their norms.
EL: Okay. And I told you, when we were emailing earlier, that you'd have to hold my hand a little bit on this. So yeah, sorry, if this is a too simple question or something. But like, what is the quadratic form like the x2 + whatever xy +whatever y2 that you get from the the Gaussian integers that you just said?
KS: Right. So if we take the Gaussian integers, if I take x+iy as a Gaussian integer, its norm is x2 + y2. That’s the form right there.
EL: Okay. Yeah. All right.
KS: And then if I were to take a subset, like all the multiples of 1+i, I'm not plugging in all x's and y's. I'm plugging in only multiples of 1+i, so you end up with a slightly different form popping out.
EL: Yeah, so I guess it's kind of like x+x then.
KK: 2x2 squared basically, right?
KS: Yeah. Yeah. You could have Yeah, various things in various different situations, but yeah.
EL: Okay, thank you. Yeah. And so, yeah, can you talk a little bit about how you encountered this theorem? If it was something that like you really loved to start out with? Or if your appreciation has grown as you have continued as a mathematician?
KS: Yeah, well, it's one of these things, so I think everybody has things that they're attracted to mathematically, they all have a mathematical personality. And there's some sort of particular kinds of things that attract you. And for me, one of the things is the sort of projection theorems that tell you that a particular structure, if you look at it a different way, has a whole different personality. And it's actually the same thing, but it has just become totally different. So I really love those things. And I've always loved number theory, because it has such simple questions. But then when you dig into them, you always run into such fascinating, complex structure hidden. And so this is one of those things that if you have that kind of personality thing you, just keep bumping into. And so for me, and all of the research that I've done and things I've been interested in, I keep coming back to this theorem and bumping into it in different places. It shows up when you study complex multiplication of elliptic curves, it shows up when you study continued fractions, it shows up all over the place. And it just seems so fundamental. And it's sort of like maybe one of the most fundamental examples of this special kind of theorem that I really enjoy.
EL: Okay.
KK: Cool.
EL: All right, well, then the next portion of the podcast is the pairing. So yeah, as you know, we like to ask our guests to pair their theorem with something that helps you appreciate the theorem even more. What have you chosen for that?
KS: So, when I think about this theorem, I just it's a treat. So I think the only thing that comes to mind really over and over again is just chocolate. I love chocolate. And that's what you should enjoy this theorem with because maybe you should just be happy enjoying it.
KK: I mean, chocolate pairs with everything.
KS: That’s true. It's a bit of a cop out.
KK: No, no, that’s okay. So our most recent favorite chocolate is Trader Joe's has this stuff. And it's got, I don’t even know what’s in it, pretzels and something else crunched up in these like bark of chocolate. And it’s a dark chocolate I really recommend it. So you must have a Trader Joe's in Boulder, right?
EL: Are you a dark, milk, or white chocolate person?
KS: Oh, definitely dark. Yeah, I have a dark chocolate problem, actually.
EL: Yeah, the Trader Joe's. For me the dark chocolate peanut butter cups are are always purchased when I go to Trader Joe’s.
KK: Dark chocolate feels healthier, right? It's got more antioxidants and a little less sugar. So you're like this is fine, less milk. Okay. All right. It's actually it's a fruit, right?
EL: It’s a bean. You’re having a black bean pate right there.
KK: That’s right.
EL: Yeah, well, Salt Lake is actually a hub of craft chocolate. We have some really wonderful chocolate makers here, like single origin, super fancy kind of stuff. So if either of you are here, we'll have to pick up some and enjoy together. And yeah, along with quadratic forms and ideal classes.
KS: Sounds wonderful.
EL: Yeah. So something I meant to talk about this earlier in the episode, but you mentioned that you'd like to illustrate things, and that is how we first met is, through mathematical illustration. So I don't know, maybe it's a failure of imagination on my part, but I always, I'm always fascinated by like, number theorists who are really into illustration as well, because I think of, like, geometry, as you know, it shapes it as the more naturally illustrate-y parts of math. But would you talk a little bit about it, you know, illustrating number theory? And if if you've done anything related to this particular theorem, or if there's something else you want to talk about with your mathematical illustration?
KS: Oh, yeah, that's a that's a great idea. Yeah. So there's actually building up gradually a wonderful community of people who are interested in illustrating mathematics. And so that's maybe one of the things that you could add a link for is the website for the community.
EL: Definitely.
KS: Yeah. And so I've always found that the way I think about mathematics is very visual. I mean, I think as human beings, we have access to this whole facility for visual thinking, because we're embedded in this three-dimensional world that we're living in. And another way that we think about mathematics, I think often, is we're using another one of our natural facilities, which is our sort of social understanding facility, where we imagine characters interacting with each other and having motivations and stuff like that. But for me, it was always a very visual thing. And so even though it wasn't taught in that way, in my mind, somehow these things were always very visual things. And so I've always been really attracted to situations where you can see some hidden geometry in number theory. And with this particular theorem, there is a little bit of nice hidden geometry. I mean, the first hint of this is that when I talked about ideals in the Gaussian integers, I visualized them as a lattice.
EL: Yeah.
KS: And in all of these number rings, you can do this, you can you can think about lattices. And you're really talking about lattices, and lattices have things like shape. And you know, there's lengths and angles and stuff like that to talk about. And so one of the really cool things that you can do is you can think about, for example, with the Gaussian integers or with some other ring of interest that you can put in the plane like this, into the complex plane, then you can ask this question, it's a natural question that people ask: how can I study the collection of objects instead of the individual objects themselves? So if you want to study the collection of lattices, say, two-dimensional lattices in the plane, then one way to do it would be okay, how do I decide on a lattice? Well, I have one vector that's generating it, and then another vector that's generating it. So let's put the first one, let's sort of ignore issues of scaling and rotation, let's put the first one down pointing from like zero to one. And then the other one is somewhere, but now you don't have any choice anymore. No more freedom. And so you can think of the plane itself as a sort of moduli space, as a parameter space for the collection of lattices. And this space has a lot of beautiful properties. So you might as well order your vectors so that we're just talking about the upper half plane. So the first vector is from zero to one, and the other one is an angle less than 180 degrees from that. And so when you start looking at the geometry of this plane, and you want to talk about well, okay, I want to look at lattices, but maybe I don't care so much about what basis I'm using for the lattice, you start to divide the plane up in certain ways. And what you discover is that the natural way to talk about this plane is using hyperbolic geometry, actually. And so all of a sudden, you're doing hyperbolic geometry. And I find myself doing hyperbolic geometry sometimes when I'm doing number theory, because when I want to look at these these objects and stuff, that's just the natural world in which they live. I mean, the mathematics kind of tells you what you have to do you. You know?
EL: Yeah.
KS: And so those are moments that I really enjoy, because you're doing something that you think is just some algebra, but all of a sudden, it turns out it's geometry.
EL: Very cool. Yeah. So we will include a link to your website, which I know has some of the cool illustrations that you've done available there. And to the illustrating math, there's an online seminar that meets monthly that is really nice to go to, if you can. And, yeah, it's a lot of fun. And I, yeah, just so many different fields of math represented with that in ways that I never would have guessed.
KS: That’s true. That's one of the nice things about that community is that there's people from all different areas that you wouldn't normally interact with, because usually you have a pretty narrow research circle, if you're doing research in mathematics. But there, you're talking to everybody. And it has a much more creative feel for that reason, you get surprised by people's ideas, because they come from just a little bit farther from your home base, you know?
EL: Yeah. And I think it also kind of pushes people to really think about how they're explaining things, where you have a shorthand when you're working with someone who is in, or talking to someone who's in such a close field, and since you don't necessarily have that same common background, people, I think, it seems like are very thoughtful about how they describe things and what they assume that you already know.
KS: Yeah, exactly. It's just good to get out of your little corner.
EL: Yeah.
KK: All right. Well, this has been great. I definitely learned something today. I did not know this connection between ideal classes and quadratic forms.
KS: Oh, I thought of one more, one book I'd like to plug.
KK: Okay. Please do.
EL: Great. Yes.
KS: Yeah, so Martin Weissman has written a book called — I'm going to get the title slightly wrong. It's An Illustrated Theory of Numbers, maybe? Oh, you have it. Oh, I got it right.
EL: Yeah. An Illustrated Theory of Numbers. It's been holding up my laptop, after I read it, I will say.
KS: Yeah, and so you were asking about illustrating number theory, and this is just a beautiful book. It's completely accessible. I used it when I was teaching an introduction in number theory for undergraduates. But it doesn't require any particular background because it starts from, you know, we’ve got the integers, we’ve got addition, we’ve got multiplication, let's do some stuff. And and he really looks hard for ways to turn theorems which are usually completely algebraic into something visual, and they're just lovely.
EL: Yeah, and really amazing illustrations, and full color, like everywhere, which I know is more expensive to make books, and that's why books so often have the color in the middle, like in a little section and not the whole thing. But I do think this is just much more pleasurable to read because it it is does use that aspect, too. And it's not as stark as every page being black and white.
KS: Yeah, it's so inviting. It's a wonderful book.
EL: Yeah. Great recommendation. Thank you so much for joining us. I really enjoyed talking with you.
KS: Yeah, me too. Thank you so much for having me on.
[outro]

For this episode, we were excited to talk to Kate Stange from the University of Colorado, Boulder about the bijection between quadratic forms and ideal classes. Below are some links you might find interesting as you listen.
Stange's website
The Illustrating Mathematics website and seminar, which meets monthly on the second Friday
An Illustrated Theory of Numbers by Martin Weissman
The Buff Classic bike ride in Boulder