[Paper Review] Fermionic forms and quiver varieties

Sergey Mozgovoy|ArXiv.org|Oct 2, 2006

Algebraic structures and combinatorial models26 references16 citations

TL;DR

This paper establishes a precise connection between fermionic forms and the Poincaré polynomials of Nakajima's quiver varieties for finite quivers of type ADE. Using combinatorial generating functions and properties of the Weyl group action on cohomology, it proves a $q$-analog of the Kirillov-Reshetikhin conjecture, confirming a version of Lusztig's fermionic conjecture by relating the generating series of Poincaré polynomials to fermionic forms involving Verma modules.

ABSTRACT

We prove a formula relating the fermionic forms and the Poincare polynomials of quiver varieties associated to a finite quiver. Applied to quivers of type ADE, our result implies a version of the fermionic conjecture of Lusztig.

Motivation & Objective

• To establish a precise relationship between fermionic forms and the Poincaré polynomials of quiver varieties associated to finite quivers.
• To prove a $q$-analog of the Kirillov-Reshetikhin conjecture for quivers of type ADE.
• To verify a modified version of Lusztig's fermionic conjecture using Verma modules instead of irreducible modules.
• To demonstrate that the fermionic forms $n(\nu,\lambda,q)$ satisfy a Weyl group symmetry property: $n(\nu,w\cdot\lambda,q) = (-1)^{l(w)}n(\nu,\lambda,q)$.
• To provide a combinatorial framework using generating functions and $q$-binomial coefficients to relate representation-theoretic invariants to geometric invariants of quiver varieties.

Proposed method

• The paper introduces fermionic forms $m(\nu,\lambda,q)$ and $n(\nu,\lambda,q)$ using $q$-binomial coefficients and root lattice data, with $n(\nu,\lambda,q)$ defined without positivity constraints.
• It employs Hausel's formula for the generating function of Poincaré polynomials of quiver varieties, expressed as a formal power series in $x_i$ and $y_i$.
• A key combinatorial identity is proven: $s(\nu) = s \cdot S_\nu \overline{s}$, where $s$ is a generating function and $S_\nu$ is a shift operator, linking the fermionic forms to the geometry of quiver varieties.
• The proof uses ring homomorphisms $\Phi_\nu$ to relate the generating functions $s(\nu)$ and $r(\nu)$ to the fermionic forms $n(\nu,q^{-1})$.
• It applies the Weyl group action on cohomology of quiver varieties to derive the symmetry $n(\nu,w\cdot\lambda,q) = (-1)^{l(w)}n(\nu,\lambda,q)$.
• The final step uses the fact that for ADE quivers, the $a$-polynomial $a(q)$ is independent of $q$, leading to $r(0,q)r(0,q^{-1}) = 1/m$, where $m$ is the product over positive roots.

Experimental results

Research questions

• RQ1 How are the fermionic forms $n(\nu,\lambda,q)$ related to the Poincaré polynomials of Nakajima's quiver varieties for quivers of type ADE?
• RQ2 Does the generating function of Poincaré polynomials satisfy a fermionic form expression involving Verma modules?
• RQ3 Can the Weyl group symmetry of the fermionic forms be derived from geometric properties of quiver varieties?
• RQ4 Is there a $q$-analog of the Kirillov-Reshetikhin conjecture that holds for quiver varieties via fermionic forms?
• RQ5 What is the precise relationship between the generating functions $s(\nu)$ and $r(\nu)$ and the fermionic forms $n(\nu,\lambda,q)$?

Key findings

• The paper proves that for any $\nu \in P_+$, the generating series of Poincaré polynomials satisfies $\sum_{\alpha \in Q_+} q^{-d(\alpha,\nu)} P(\mathcal{M}(\alpha,\nu),q) e^{\nu - \alpha} = \sum_{\lambda \in P} n(\nu,\lambda,q^{-1}) \operatorname{ch} M(\lambda)$, confirming a $q$-analog of the Kirillov-Reshetikhin conjecture.
• For quivers of type ADE, the $a$-polynomial $a(q)$ is independent of $q$, so $r(0,q)r(0,q^{-1}) = 1/m$, where $m = \prod_{\alpha \in \Delta_+} (1 - x^\alpha)^{-1}$.
• The fermionic form $n(\nu,\lambda,q)$ satisfies the Weyl group symmetry $n(\nu,w\cdot\lambda,q) = (-1)^{l(w)}n(\nu,\lambda,q)$, which is consistent with the cohomological action of the Weyl group on quiver varieties.
• The identity $s(\nu) = s \cdot S_\nu \overline{s}$ is established as a key combinatorial tool linking the generating functions to the fermionic forms.
• The proof shows that $n(\nu,q^{-1}) = r(\nu,q) / (r(0,q) r(0,q^{-1}))$, and since $r(0,q)r(0,q^{-1}) = 1/m$, it follows that $n(\nu,q^{-1}) = m \cdot r(\nu,q)$, linking the fermionic form to the character of Verma modules.
• The result implies that the generating function of Poincaré polynomials is equal to $m \sum_{\lambda \in P} n(\nu,\lambda,q) e^{\lambda - \nu}$, and since $\operatorname{ch} M(\lambda) = e^\lambda m$, this yields the final identity in Theorem 1.1.