The next installment of the “Math And…” seminar is coming next week:

7 Mar 2008, 2:30 PM, Birge 145:

Dmitri Tymoczko(Princeton University, music)Title: “Geometry and Music.”

In my talk, I will explain how to translate the basic concepts of music theory into the language of contemporary geometry. I will show that musicians commonly abstract away from five types of musical transformations, the “OPTIC transformations,” to form equivalence classes of musical objects. Examples include “chord,” “chord type,” “chord progression,” “voice leading,” and “pitch class.” These equivalence classes can be represented as points in a family of singular quotient spaces, or orbifolds: for example, two-note chords live on a Mobius strip whose boundary acts like a mirror, while four-note chord-types live on a cone over the real projective plane. Understanding the structure of these spaces can help us to understand general constraints on musical style, as well as specific pieces. The talk will be accessible to non-musicians, and will exploit interactive 3D computer models that allow us to see and hear music simultaneously.

The interactive 3D computer models are made by Dmitri’s program Chord Geometries, which you can download from his website. Or you can just watch the sample movies, and see what the opening of “Smoke on the Water” looks like on a Mobius strip. Dmitri comes by his mathematical know-how familially; his younger sister is the algebraic geometer Julianna Tymoczko (with whom I wrote this paper about diameters of finite groups) and his father was the philosopher of mathematics Thomas Tymoczko.

I’ve booked a big room, and the talk is open to the public, so if the abstract sounds interesting to you, please come!

On the other hand, perhaps the abstract puts you off by means of words like “real projective plane.” Well, it’s written for mathematicians; maybe it’s worth explaining in more down-to-earth language about what a guy like Dmitri means by “the space of chords,” or, as an algebraic geometer like me might put it, *the moduli space of chords. *Explanation below the fold: it’s essentially a brief version of an article called “The Idea of a Moduli Space” that I wrote for *Imagine* magazine when I was a grad student.

**Non-definition:** Let S be some set of things you’re interested in. A *moduli space* for S is some kind of shape M with the properties that

- The points of M are in 1-1 correspondence with the elements of S;
- Two points of M are close together if and only if the corresponding elements of S are “close together,” with respect to whatever notion of “closeness” applies to S.

This is a non-definition because there are lots of words whose meanings I haven’t specified — “shape,” and “close together” to start with. But it’s good enough to start making examples. For instance, the moduli space of real numbers is a line — namely, the number line you learned about in grade school. And indeed, two numbers are close to each other (in the numerical sense) just when the corresponding points on the line are close to each other. Similarly, the moduli space of ordered pairs (x,y) of real numbers is a plane — this was Descartes’ great innovation, now so familiar that it seems almost obvious.

What about lines in the plane which pass through the origin? Well, such a line is determined by the angle it makes with the horizontal, which is some real number in the interval [0,pi). So you might say, “the moduli space is the interval [0,pi).” But that’s not right! Because 0 and 3.14159, for instance, are very far from each other in the interval, but the corresponding lines are almost identical. So condition 2 above is violated. (This might be clearer if I knew how to make pictures on the blog!) The moduli space needs to keep track of the fact that the two ends of this interval, rather than being far apart, are actually supposed to be close to each other; this can be accomplished by grabbing the interval at both ends and bending them together until they form a circle. And it is the circle that’s the moduli space for lines through the origin.

(Exercise for the energetic reader: what is the moduli space of all lines in the plane?)

What about *unordered* pairs of real numbers? Well, you might again be tempted to say “the plane” — after all, given a pair of numbers {x,y}, you just write down the point (x,y) in the Cartesian plane. But now the points (2,1) and (1,2) refer to the same unordered pair; so condition 1 is violated. We need to *identify* those two points, and indeed, to identify (x,y) with (y,x) for every pair of real number (x,y). You can envision this process geometrically as “folding the plane in half” along the line x=y.

This last example is the most relevant one for Dmitri’s talk. When we talk about a pair of real numbers, we have a choice — do we care what order they come in, or do we consider two pairs of real numbers the same if they differ only with respect to ordering? This choice is important — in the first case, the moduli space is the plane, in the second case, only the half plane.

Similarly, when studying music theory, one has choices about how to think of chords. When we say two chords are “the same,” do we mean they contain the exact same sequence of notes? Or are two chords the same if one is the transpose of the other by an octave? What if one is the transpose of the other by some other interval? What if chord 1 is chord 2 with the bottom note transposed an octave up, but the rest left alone? What if chord 1 is chord 2 upside down? Any one of these relations — and there are more in this vein — would lead some music theorists, in some contexts, to say the chords were “the same.” And every choice about which pairs of chords are “the same” leads to a different moduli space. Dmitri has worked out a really nice description of these moduli spaces, which apparently organizes and unifies lots of previous work in music theory — in particular, it provides a very natural and geometric notion of what it means for one chord to be “close” to another.

We’re delighted to have him here — I hope you can come.

The name “moduli space” really grates. Normally in English we don’t have plural components in noun compounds (other than the last noun, of course): since “moduli” is the plural of “modulus”, “moduli space” sounds as bad to me as “insects eater” instead of “insect eater” for a creature that eats insects. (There are a few exceptions, like “enemies list”, to be sure.)

Haha, it’s amusing to see a piece that you’ve played looks like on a Mobius strip.

[...] moduli space of senators and the moduli space of movies 04Mar08 Last week I blogged about Dmitri Tymoczko’s lecture and the moduli space of chords; since then I remembered some more nice examples of “moduli spaces” in the loose sense [...]

Nice works ! keep trying to interpreted !!! Because there still a lot of secret in this universe!!!

We can tell whether an hypothesis involves genuine science by asking: “What evidence would be required to disconfirm this hypothesis?”

In this case, we have some pretty looking pictures, and some math, and the implicit claim that the prettiness of the graphs somehow relates to the prettiness of the music. Alas, the prettiness of the graphs remains a subjective question which cannot be resolved with double-blind experiments…as does the prettiness of the music.

So there’s clearly no testable scientific hypothesis here. That being the case, we have something that uses the language of science (mathematics) and looks like science (plenty of equations and scientific jargon), but isn’t actually science. What do we call something like that?

Three guesses.

Music theory boasts a long history of such escapades. You might find intriguing Sorbonne Professor Laurent Fichet’s 1996 book

, Libraire J. Vrin, Paris, 1996. (Scientific Theories of the 19th and 20th centuries.) Alas, it’s only available in French, but well worth reading. The review by David Perrott hits the high points.Les théories scientifiques de la musique: XIXčme et XXčme sičcles

Oh please, Mclaren, that’s hardly a serious criticism unless you’re trying to suggest that because Tymoczko published in the journal Science, he’s therefore trying to pass his work off as science.

The language of physics may be mathematics, but most of the rest of the sciences are only marginally mathematical. Whereas accountants and financiers speak the language of mathematics, without making any claims to practicing “science.” Most significantly, the people who speak the language of mathematics most fluently, i.e. mathematicians, would seldom claim to be scientists in the Popperian sense.

Famously, even Popper’s own criteria of falsifiability is not falsifiable, and therefore not scientific by it’s own measure. These distinctions are not hard and fast, and letting yourself become too encumbered by them only diminishes your soul. Tymoczko is not claiming that he’s achieved cold-fusion, after all. Music, however, has long enjoyed a rich and appropriate relationship with mathematics, and both have benefited (even science did if you’ve ever studied Kepler), without either ever worrying about whether they were following the appropriate experimental protocal.

What Tymoczko seems to be presenting is a model that gives a deeper, geometric explanation for certain types of regularities that seem to arise in the musical that humans have often found pleasurable. That’s interesting. Pythagoras was said to have discovered the division of the scale by listening to the notes, and finding ratios. Is it falsifiable? No. Did we eventually discover that Pythagoras’ scale possessed deeper mathematical, and even physical significance than first thought? Absolutely. In fact, it’s intrinsically involved in our basic understanding of wave phenomena, as I’m sure you know.

So maybe you should trust your ears, investigate the geometries, and experience the wonder that these sorts of connections can hold. But get out of the criticism business–it’s neither good science nor good philosophy, at least not as you’re practicing it.

“Popperian sense” is nonsense. Philosophers make me laugh sometimes, they think in circles.

“Fourth, one could investigate how distances in the orbifolds relate to perceptual judgments of chord similarity.”

If mclaren had read the paper that he attacked, perhaps he’d have noticed that the suggestions for further research asked for empirical psychological studies of the variety that Popper fought for (against the then popular Freudian Psychoanalysis).

Though Dan is completely and unambiguously right about the falsificationism of falsificationism, I’d like to point out that Popper was very clear that he was founding it as an Epistemology on top of a particular Ontology.