Abstract
The EA-matrix integrals, introduced in Barannikov (2006), are studied in the case of graded associative algebras with odd or even scalar product. This work proves that the EA-matrix integrals for associative algebras with scalar product are integrals of equivariantly closed differential forms with respect to the Lie algebra $gl_{N}(A)$.
Introduction
In classical differential geometry, manifolds are characterized by their topology and calculus. In the realm of noncommutative geometry, researchers study "noncommutative varieties"—abstract spaces where the coordinates do not necessarily commute ($xy \neq yx$). A fundamental challenge in this field is defining and calculating "periods" for these spaces. In classical geometry, periods correspond to volumes or integrals of differential forms; in the noncommutative setting, they are essential for understanding the underlying structure of the variety.
Previously, EA-matrix integrals were introduced as a mechanism to compute these periods and relate them to Gromov-Witten invariants, which count curves on geometric spaces. However, direct computation of these integrals is notoriously difficult. This paper provides a robust geometric framework by demonstrating that these integrals are not merely formal sums but are deeply rooted in the principle of equivariant localization.
Theoretical Framework: BV Formalism and Equivariant Cohomology
The study focuses on graded associative algebras equipped with either even or odd scalar products. To analyze the matrix integrals, the paper employs several sophisticated mathematical tools:
- Batalin-Vilkovisky (BV) Formalism: Originally a tool for gauge theories in physics, the BV differential ($\Delta$) is used here to define "closed" elements within the algebra of functions on matrix spaces.
- Equivariant De Rham Cohomology: The matrix integral is treated as a differential form that remains "closed" under the action of a specific symmetry group—the Lie algebra $gl_N(A)$.
A critical requirement for the stability of these integrals is the unimodularity of the underlying Lie algebra $A_0$. Unimodularity (where the trace of the adjoint action is zero) ensures that the volume element remains invariant under the symmetry action, allowing the geometric properties to be preserved during integration.
The Core Insight: Equivariant Localization
The central contribution of this paper is the realization that the integrand of an EA-matrix integral is an equivariantly closed form. This discovery allows the application of the Equivariant Localization Principle.
In mathematics, localization suggests that the integral of a symmetric form over a high-dimensional space can be reduced to a much simpler calculation over the "fixed points" of that symmetry. By applying this to EA-matrix integrals, the author shows that complex, high-dimensional matrix calculations can be simplified into expressions involving determinants. This reduction makes the integrals significantly more manageable and reveals their internal symmetry.
Key Results and Connections
The paper establishes several landmark results that bridge noncommutative geometry and mathematical physics:
- Theorem 1: The author formally proves that the product of the closed De Rham differential form and the exponential function used in the EA-matrix integral is equivariantly closed with respect to the $gl_N(A)$ action. This establishes a direct correspondence between the Batalin-Vilkovisky differential and the classical De Rham differential.
- Integrable Hierarchies: When expressed in terms of specific variables, these integrals are shown to be $\tau$-functions of the KP-type hierarchy. This links noncommutative geometry to the theory of solitons and integrable systems, as the integrals satisfy the Hirota quadratic equations.
- Ribbon Graph Expansion: The results reinforce the connection between EA-matrix integrals and the summation over "stable ribbon graphs." These graphs are instrumental in calculating cohomology classes on the moduli spaces of curves, providing a combinatorial bridge to algebraic geometry.
Conclusion and Future Directions
By framing EA-matrix integrals as equivariantly closed forms, this work simplifies the computation of periods in noncommutative geometry and integrates them into the broader landscape of mathematical physics. While the current focus is on associative algebras, the author notes that these results likely extend to $A_\infty$-algebras and categories, which are central to modern mirror symmetry.
Future research may address the constraints of the scalar product. While some results assume a positive definite scalar product, the author suggests that "partial Wick rotations"—a technique borrowed from physics—could handle non-definite signatures, though the explicit calculation of specific examples via determinants remains a subject for further specialized study.
Summary updated