Table of Contents
Fetching ...

Generating functions for intersection numbers on moduli spaces of curves

Andrei Okounkov

TL;DR

The paper establishes a precise, multivariate error-function representation for the n-point functions of psi-class intersection numbers on moduli spaces of curves, revealing a deep link between moduli-space intersection theory and edge scaling in the GUE matrix model. It derives a closed-form formula for the n-point function in terms of an integral function $\mathcal{E}$, its cyclic symmetrization, and a Möbius-connected combination, with the final expression expressed as a symmetric function of $x_i$ and aligned with a GKZ–hypergeometric framework. By connecting Laplace transforms of the Airy kernel to Kontsevich’s ribbon-graph expansion, the paper also provides an alternative path to Witten’s KdV equations via higher Fay identities (ASV), identifying the relevant KdV $\tau$-function and demonstrating how its coefficients reproduce the intersection-number hierarchy. The work highlights cancellations and selection rules that simplify the asymptotics, clarifying which terms contribute to the connected, positive-degree part and establishing a rigorous bridge between random-matrix edge phenomena and the KdV integrable hierarchy with potential applications to moduli-space enumerative geometry. Overall, the results offer a unified viewpoint on n-point functions, map asymptotics, and integrable structures in a single matrix-model framework.

Abstract

Using the connection between intersection theory on the Deligne-Mumford spaces and the edge scaling of the GUE matrix model (see math.CO/9903176, math.AG/0101147), we express the n-point functions for the intersection numbers as n-dimensional error-function-type integrals and also give a derivation of Witten's KdV equations using the higher Fay identities of Adler, Shiota, and van Moerbeke.

Generating functions for intersection numbers on moduli spaces of curves

TL;DR

The paper establishes a precise, multivariate error-function representation for the n-point functions of psi-class intersection numbers on moduli spaces of curves, revealing a deep link between moduli-space intersection theory and edge scaling in the GUE matrix model. It derives a closed-form formula for the n-point function in terms of an integral function , its cyclic symmetrization, and a Möbius-connected combination, with the final expression expressed as a symmetric function of and aligned with a GKZ–hypergeometric framework. By connecting Laplace transforms of the Airy kernel to Kontsevich’s ribbon-graph expansion, the paper also provides an alternative path to Witten’s KdV equations via higher Fay identities (ASV), identifying the relevant KdV -function and demonstrating how its coefficients reproduce the intersection-number hierarchy. The work highlights cancellations and selection rules that simplify the asymptotics, clarifying which terms contribute to the connected, positive-degree part and establishing a rigorous bridge between random-matrix edge phenomena and the KdV integrable hierarchy with potential applications to moduli-space enumerative geometry. Overall, the results offer a unified viewpoint on n-point functions, map asymptotics, and integrable structures in a single matrix-model framework.

Abstract

Using the connection between intersection theory on the Deligne-Mumford spaces and the edge scaling of the GUE matrix model (see math.CO/9903176, math.AG/0101147), we express the n-point functions for the intersection numbers as n-dimensional error-function-type integrals and also give a derivation of Witten's KdV equations using the higher Fay identities of Adler, Shiota, and van Moerbeke.

Paper Structure

This paper contains 12 sections, 6 theorems, 114 equations, 2 figures.

Key Result

Theorem 1

Figures (2)

  • Figure 1: A typical map of large perimeter
  • Figure 2: A piecewise linear function $\gamma$

Theorems & Definitions (9)

  • Definition 1.1
  • Theorem 1
  • Proposition 2.1
  • Proposition 2.2
  • Definition 2.3
  • Proposition 2.4
  • Proposition 2.5
  • Lemma 2.6
  • proof