[Paper Review] Geometric Proof of a Conjecture of Fulton

Prakash Belkale|ArXiv.org|Nov 28, 2005

Advanced Algebra and Geometry4 references4 citations

TL;DR

This paper provides a geometric proof of Fulton's conjecture on the multiplicities of irreducible representations in tensor products of GL(r) representations, using geometric invariant theory (GIT) moduli spaces and Schubert calculus. The key result confirms that the Littlewood-Richardson coefficient $ c^{ u}_{ u, u} = 1 $ if and only if $ c^{N u}_{Neta,Neta} = 1 $ for all positive integers $ N $, establishing a deep stability property of tensor product multiplicities.

ABSTRACT

We give a geometric proof of a conjecture of W. Fulton on the multiplicities of irreducible representations in a tensor product of irreducible representations for GL(r).

Motivation & Objective

• To provide a geometric proof of Fulton's conjecture on the stability of Littlewood-Richardson coefficients under scaling of weights.
• To establish a categorical framework based on GIT moduli spaces that generalizes beyond the original conjecture.
• To demonstrate that the vanishing and non-vanishing of tensor product multiplicities are preserved under scaling, using geometric intersection theory.
• To lay the foundation for generalizations to quantum cohomology and quiver representations.

Proposed method

• The proof uses Schubert varieties and their intersections in Grassmannians to model tensor product multiplicities via geometric invariant theory (GIT).
• It applies the tangent space method to analyze the dimension of intersections of Schubert cells in flag varieties, using the expected dimension formula involving codimensions of Schubert varieties.
• The construction involves filtrations of subspaces and graded quotients with specified inclusions into quotient spaces, ensuring compatibility with flag structures.
• A key technique is the use of the rank of the space $ \operatorname{Hom}_{\mathcal{I}}(V,Q,\mathcal{F},\mathcal{G}) $, which determines the dimension of the intersection of Schubert varieties.
• The proof relies on Kleiman’s transversality theorem to ensure genericity and irreducibility of the intersection loci.
• The argument reduces the conjecture to a dimension count on tangent spaces and uses a filtration construction to verify the expected dimension matches the actual dimension.

Experimental results

Research questions

• RQ1 Does the condition $ c^{\lambda}_{\mu,\nu} = 1 $ imply $ c^{N\lambda}_{N\mu,N\nu} = 1 $ for all positive integers $ N $?
• RQ2 Can the stability of tensor product multiplicities be proven geometrically using Schubert calculus and GIT?
• RQ3 Is the dimension of the intersection of Schubert varieties in flag manifolds determined by a canonical filtration and graded maps?
• RQ4 Can the geometric approach be extended to quantum cohomology and quiver representations?

Key findings

• The conjecture is confirmed: $ c^{\lambda}_{\mu,\nu} = 1 $ if and only if $ c^{N\lambda}_{N\mu,N\nu} = 1 $ for all $ N \geq 1 $, via a geometric argument based on GIT moduli spaces.
• The dimension of the intersection of Schubert cells is computed as $ r(n-r) - \sum_{j=1}^{s} \sum_{a=1}^{r} (n-r+a - i^{j}_a) $, which matches the expected dimension when the intersection is non-empty.
• The existence of a filtration $ S^{(h)} \subset \cdots \subset S^{(0)} = V $ with compatible maps into $ Q $ ensures that the dimension of the Hom space matches the expected dimension.
• The space $ \operatorname{Hom}_{\mathcal{I}}(V,Q,\mathcal{F},\mathcal{G}) $ has rank equal to the expected dimension plus a correction term involving the highest-level subspace $ S^{(h)} $.
• The proof establishes that the intersection $ \cap_{j=1}^{s} \Omega^{o}_{I^{j}}(E^{j}_{\bullet}) $ is non-empty if and only if the dimension of the Hom space equals the expected dimension.
• The geometric method provides a categorical framework suitable for generalization to quantum cohomology and quiver representations.