I would like to compute the Hilbert series of the ring of invariants of certain irreducible representations of some groups (namely $SO(5)$ to begin with).

To put it in some broader context, let $G$ be a simple complex Lie group with compact form $K$ and $V$ an irreducible representation of $G$. Choose $T \simeq (\mathbb{C}^*)^l$ a maximal torus of $G$, with $l$ being the rank of $G$.

The (excellent) book Computational invariant theory by H. Derksen and G. Kemper suggests the following strategy:

Let $((\mu_i, m_i)_{i \in I}$ be the list of weights/multiplicities of $V^*$, with $\mu_i: T \to \mathbb{C}^*$.

Define $$ \chi_0(t) = \sum_{k=0}^\infty t^k \chi^{\mathrm{Sym}_k(V^*)} $$ with $\mathrm{Sym}_k(V^*)$ the $k$-th symmetric tensor power of $V^*$, i.e. the space of homogeneous polynomials of degree $k$ on $V$. Then it is not complicated to see that $$ \chi_0(t, z) = \prod_{i \in I} \frac{1}{\left(1 - t z^{\mu_i}\right)^{m_i}}, $$ where $z^\mu_i = \mu_i(z)$.

Let $D = \prod_{\alpha \in \Phi^+} (1 - z^{-\alpha})$ be the Weyl denominator. Then (Corollary 4.6.9) the Hilbert series $H(\mathbb{C}[V]^{G}, t)$ of $\mathbb{C}[V]^{G}$ (the subring of invariant polynomials) is given by the constant coefficient in $$ \chi_0(t, z) D = \frac{\prod_{\alpha \in \Phi^+} (1 - z^{-\alpha})}{\prod_{i \in I} \left(1 - t z^{\mu_i}\right)^{m_i}} $$ seen as an element in $\mathbb{C}(t)[z_1, z_1^{-1}, \ldots, z_l, z_l^{-1}]$. It follows from the residue formula that we have $$ H(\mathbb{C}[V]^{G}, t) = \frac{1}{(2\pi i)^l}\int_{\mathbb{U}^l} \chi_0(t, z) D \frac{dz_1}{z_1} \ldots \frac{dz_l}{z_l}, $$ where $\mathbb{U} = \{z \in \mathbb{C}, |z|=1\}$.

The reasoning is clear to me but, when I put it on a computer and try to do explicit calculations for rank $2$ Lie groups things get nasty. Calculations work for $SO(4)$ (because of the isomorphism $\widetilde{SO}(4) = SU(2) \times SU(2)$) but already for $SO(5)$ the calculation is intractable (I have been using Sage). The representation I would be interested in is the one with highest weight $2 \overline{\omega}_2$, namely, the Cartan product of $$ \Lambda_2 \mathbb{C}^5 $$ with itself.

Is there any known modification to this formula and/or fast algorithm to compute the integral above?

In particular, is there any known mean to take advantage of the Weyl group of $G$? The integrand above is not invariant under the Weyl group but we could imagine computing $$ \int_K \chi_0(t, g) d\mu(g) $$ where $\mu$ is the Haar measure of $K$ but means of the Weyl integration formula.

Any help would be invaluable.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.