Formal Summation and Dirichlet L-functions

Recall the classical Riemann zeta function:

\zeta(s) = \sum_{n\geq 1}\frac{1}{n^s}

and the Dirichlet L-functions for a character \chi: \mathbb Z \to \mathbb C:

S(s,\chi) = \sum_{n\geq 1}\frac{\chi(n)}{n^s}.

defined for \Re(s) > 1.  These functions can be analytically continued to the entire complex plane (except a pole at s=1 in the case of \zeta(s)). In particular, the values at non positive integers carry great arithmetic significance and enjoy many properties.

For instance, L(s,\chi) is an algebraic integer and in fact equal to B_{k+1,\chi}/(k+1) where B_{k,\chi} are the generalized Bernoulli numbers. Moreover, these values satisfy p-adic congruences and integrality properties such as the Kummer congruence (and generalizations to the L-functions).

The standard proof of these proceeds by showing that L(s,\chi) satisfies a functional equation that relates L(1-s,\chi) to L(s,\overline{\chi}) and then computing the values L(n,\chi) for n \geq 1 integral using analytic techniques (such as Fourier analysis).

Proving the p-adic properties is then by working directly with the definition of the generalized Bernoulli numbers instead of the L-functions. However, the L-functions are clearly the fundamental object here and it would be nice to have a way to directly work with the values at negative integers (without using the functional equation).

One might be tempted to extend the series definition to the negative integers and say:

\zeta(-k)  "="  1^k + 2^k + 3^k + \dots

For instance, consider the following (bogus) computation:

\zeta(0)\frac{t^0}{0!} = 1^0\frac{t^0}{0!} + 2^0\frac{t^0}{0!} + \dots

\zeta(-1)\frac{t^1}{1!} = 1^1\frac{t^1}{1!} + 2^0\frac{t^1}{1!} + \dots

\zeta(-2)\frac{t^2}{2!} = 1^2\frac{t^2}{2!} + 2^2\frac{t^2}{2!} + \dots

. . .

and let us “sum” the columns first:

\sum_{k\geq 0}\zeta(-k)\frac{t^{k}}{k!} = e^{t} + e^{2t} + \dots = \frac{e^t}{1 + e^t}

which, remarkably enough, is the right generating function for \zeta(-k)! We have exchanged the summation over two divergent summations and ended up with the right answer.

In fact, it is possible to rigorously justify this procedure of divergent summation and moreover, one can use it to prove a lot of arithmetic properties of these values rather easily (like the Kummer congruence). I learnt the basic method from some lecture notes of Prof. Akshay Venkatesh here: Section 3, Analytic Class Number formula and L-functions.

I then discovered that one could use these techniques to compute the explicit values (along the outline above) and prove some more stuff. I wrote this up in an article (that also explains the basic technique and should be (almost) self contained) here: Divergent_series_summation.

The Groupoid Cardinality of Finite Semi-Simple Algebras

A groupoid is category where all the morphisms are isomorphisms and groupoid cardinality is a way to assign a notion of size to groupoids. Roughly, the idea is that one should weigh an object inversely by the number of automorphisms it has (and we only count each isomorphic object as one object).

It is important to count only one object from each isomorphism class since we want the notion of groupoid cardinality to be invariant under equivalences of groupoids (in the sense of category theory) and every category is equivalent to it’s skeleton. For further motivation for the idea of a groupoid cardinality, see Qiaochu Yuan’s post on them.

This seems like quite a strange thing to do but it turns out to be quite a useful notion. One of my favorite facts about Elliptic curves is that the groupoid cardinality of the supersingular elliptic curves in characteristic p is p-1/24! See the Eichler-Deuring mass formula. 

Another interesting computation along these lines is that the number of finite sets is e. One can ask this question of various groupoids and the answer is often interesting. I will ask it today of semi-simple finite algebras of order n. By an algebra, I will always implicitly mean commutative in this post.

 

Continue reading

A nice proof that the irreducible characters form a basis for class functions

 

Warning: This is a very shoddily organized post and can be vastly improved. The method of proof is still nice however.

Let G be a finite group of size g and k an algebraically closed field such that g \neq 0 in it. Let V_1,\dots, V_n be the irreducible representations of G over k and \chi_1,\dots,\chi_n the corresponding characters.

A class function f: G \to k is a set function such that f is constant on conjugacy classes. That is, for all g,h \in G, f(h^{-1}gh) = f(g). It is easy to see that any character is a class function by cyclicity of the trace function.

Let us denote by H the vector space of class functions and W the vector space of functions spanned by the \chi_i within H. It is natural to wonder about the size of W relative to H. In fact, the two vector spaces are equal!

To prove this, let us consider the algebra k[G]. By our hypothesis and the usual argument of Maschke’s theorem, this is a semisimple ring. That is, k[G] is semisimple as a G representation. Let r_G be the character attached to this so called regular representation. A simple computation shows us that r_1 = g and r_G = 0 otherwise.

Moreover, for any irreducible representation V_k, we see from last time that:

Hom_G(V_k,k[G]) = \frac{1}{g}\sum_g r_G(g^{-1})\chi_k(g) = \chi_k(1) = \dim V_k.

Therefore, by semisimplicity

k[G] \cong \oplus_k V_k^{\dim V_k}

as representations. Moreover, the orthogonality relations from last time also show that the \chi_k are linearly independent. That is, \dim W = \# irreducible characters.

Taking the endomorphism ring on both sides (as modules over k[G]), we obtain:

k[G] \cong M_{\dim V_k}(k)

as algebras. Considering dimensions, we have incidentally shown that g = \sum_k \dim V_k^2. Moreover, let us consider the central elements in k[G]. They are easily seen to exactly be the elements of the form \sum_g f(g)g for $f:G \to k$ a class function.

On the other hand, the central elements in M_n(k) are always just k. Therefore, the dimension of the center of k[G] (= vector space of class functions) is equal to the number of irreducible representations. Therefore, \dim W = \dim H and we are done.

The above map can be thought of as the representation of a class function in the basis defined by the functions c_k\overline{\chi_k} for some appropriate constants c_k. This is because the element:

\lambda_i = \sum_{g}\overline{\chi_i}g \in k[G]

maps to a a diagonal element in M_{\dim V_j}(k) that can be calculated by taking the trace. That is to say, the image of \lambda_i in $latex M_{\dim V_j}(k) is given by:

\frac{1}{\dim V_k}\sum_{g}\overline{\chi_i(g)}\chi_j(g) = \frac{g}{\dim V_k}\delta_{ij}

by the orthogonality relations. We see that c_k = \frac{\dim V_k}{g}.