Basic Notions Seminar talk on October 11, 2002

Speaker:   Arkady Berenstein (University of Oregon)

Title:   Macdonald's Identities

Abstract: 
In 1748 Leonard Euler found a beautiful power series expansion for the following infinite product:

phi(x)=(1-x)(1-x^2)(1-x^3)...

Surprisingly enough, after multiplying out, almost all powers of x got canceled!

    Seventy years later Karl-Friederich Gauss raised the same question about the cube of phi(x), and the power series expansion turned out to be even more beautiful. Gauss' result is especially remarkable because there is nothing special about the square of phi(x).
    Since then, many mathematicians were puzzled by existence of certain exponents n for which phi(x)^n has a nice power series expansion with lots of cancelations: the case n=8 was studied by Klein, n=14, 26 by Atkin, n=24 by Ramanujan, and so on.
    Apparently, the most comprehensive list of such "good" exponents was obtained by Freeman Dyson in late sixties. Here is the beginning of the list:
n=1, 3, 8, 10, 14, 15, 21, 24, 26, 28, 35, 36, ...

Dyson tried very hard to understand the origin of these numbers but with no success.
The real breaktrough in computing phi(x)^n was due to Yan Macdonald (who worked at Princeton at the same time as Dyson). Macdonals noticed that:
    - The good exponents n are dimensions of simple Lie algebras;
    - The identity for phi(x)^n is a very special case of the Denominator Identity for Kac-Moody Lie algebras. The Denominator Identity itself is a generalization of the Wandermond determinant identity.
In my talk I will describe Macdonald identities and will outline their relation to the Kac-Moody theory.