For partitions $\lambda, \mu \vdash n$, the Kostka-Foulkes polynomial $K_{\lambda,\mu}(q)$, a $q$-analog of the Kostka coefficient $K_{\lambda,\mu}$, has a combinatorial description, due to Lascoux and Schützenberger, as $$ K_{\lambda,\mu}(q) = \sum_{T} q^{\mathrm{charge}(T)},$$ where the sum is over semistandard Young tableaux of shape $\lambda$ and content $\mu$. In the case of standard tableaux, charge is basically the same as major index (maybe we need to consider cocharge instead); at any rate, what is true is that if $\mu=(1,1,\ldots,1)$, then we have the $q$-hook length formula (due to Stanley) which gives a product formula for $K_{\lambda,\mu}(q)$: $$ K_{\lambda,\mu}(q) = \sum_{\textrm{$T$ a SYT of shape $\lambda$}} q^{\mathrm{maj}(T)}= \frac{[n]!_q}{\prod_{u\in\lambda} [h(u)]_q}.$$

The Kostka-Foulkes polynomials are the Type A version of Lusztig's $q$-analog of weight multiplicity (see papers of Lusztig and Joseph-Letzter-Zelikson cited below for the definition of these). As I understand it, giving a combinatorial description of these in other types as a statistical generating function over tableaux analogous to the charge formula is a pretty difficult problem. This problem is not fully resolved but the theory of crystals has led to significant progress (see the recent papers of Lecouvey-Lenart https://arxiv.org/abs/1707.03314 and Jang-Kwon https://arxiv.org/abs/1908.11041).

My question however is not about statistical generating functions, however. It is rather about product formulas.

Question: Are there any examples of Lusztig's $q$-analog of weight multiplicity, beyond the SYT $q$-hook length formula explained above, for which we have a product formula?

(Note that when we pass to the more general root system formulation, content $\mu=(1,1,\ldots,1)$ actually corresponds to weight $\mu=0$ since we mod out by $(1,1,\ldots,1)$ to form the weight lattice in Type A. But also note that not every Type A $q$-analog of weight multiplicity of the form $K_{\lambda,0}(q)$ has a hook-length formula because we also need that $\lambda$ is a partition with the right number of boxes.)

Lusztig, George, Singularities, character formulas, and a (q)-analog of weight multiplicities, Astérisque 101-102, 208-229 (1983). ZBL0561.22013.

Joseph, Anthony; Letzter, Gail; Zelikson, Shmuel, On the Brylinski-Kostant filtration, J. Am. Math. Soc. 13, No. 4, 945-970 (2000). ZBL0991.17006.

EDIT: I just realized that the question still makes sense even removing the "q" from it. That is...

q=1 version of question: Are there other product formulas for weight multiplicities of simple Lie algebras, beyond the Type A cases corresponding to SYTs?

  • 2
    $\begingroup$ the q-analogue of $0$-weight multiplicity of an irreducible $\lambda$-highest weight representation of $SL_n$ (where $\lambda$ is a partition into at most $n$ parts) is the Kostka-Foulkes polynomial $K_{\lambda,\mu}(q)$ where $\mu=(k^n)$ when $kn=|\lambda|$ - this one does not seem to have a product formula - not even palindromic $\endgroup$ Jul 28, 2023 at 9:09

2 Answers 2


It looks like one has a product formula for the Poincare series of generalized exponents (Lusztig's $q$-multiplicity for the $0$ weight) of small dominant weights, i.e. those dominant weights in the root lattice for which twice a dominant root does not appear in the corresponding highest weight representation

two sets of evidences:

  1. the Stanley result is capturing the generalised exponents of type A small dominant weights

  2. all examples computed at the end of section 1 in types B,C and D in https://link.springer.com/article/10.1007/s40574-023-00390-8


For some skew shapes, there might be product formulas, I suggest to look at the Pak-Panova-Morales series of papers.

When it comes to major-index generating functions of posets, there is a q-analog proved for trees and forests, due to A. Bjorner and M. Wachs, which has a nice product formula. There is also the class of D-complete posets, which allow for product formulas.

For shifted shapes, see e.g. this document.

Although, the connection with charge is perhaps not clear.

Coincidentally, I am currently working on a project which will answer some of your questions regarding charge for certain Kostka-Foulkes polynomials...

EDIT: For things closer to weights, see e.g. this twisted version of Kostka-Foulkes polynomials, which gives weight multiplicities of tensor products of twisted representations into twisted representations (it seems).

  • $\begingroup$ Are any of these actually $q$-analogs of weight multiplicity? $\endgroup$ Aug 30, 2019 at 9:34
  • 1
    $\begingroup$ I think maybe my question was unclear. It was intended as: is there a root system $\Phi$ and choice of weights $\lambda$, $\mu$, such that we have a product formula for Lusztig's $q$-analog of weight multiplicity $K_{\lambda,\mu}(q)$, beyond $\Phi=A_n$ and the $\lambda$,$\mu$ corresponding to Standard Young Tableaux (for which the $q$-hook length formula applies)? $\endgroup$ Aug 30, 2019 at 19:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.