Abstract:The Berezin-Karpelevich integral is a double integral over unitary matrices which plays the role of the Itzykson-Zuber integral in rectangular matrix models. We obtain a topological expansion of the Berezin-Karpelevich integral in terms of monotone Hurwitz numbers, and obtain from this certain combinatorial identities.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the topological expansion problem of the Berezin - Karpelevich integral. Specifically, the goal of the paper is to establish the topological expansion of the string coefficients and the connected string coefficients of the Berezin - Karpelevich integral, and these coefficients can be represented by the combinatorial refinement of the monotone double Hurwitz numbers.
### Background Introduction
The Berezin - Karpelevich integral is a double integral over the unitary matrix, and it plays a role in the rectangular random matrix model similar to that of the Itzykson - Zuber integral. The Itzykson - Zuber integral has been widely studied in many research projects, especially in the multi - matrix model. This integral depends on a coupling parameter \(z\in\mathbb{C}\) and a pair of matrices \(A, B\in\mathbb{C}^{N\times N}\), and its dependence is only reflected through the eigenvalues \(a_1,\ldots,a_N,b_1,\ldots,b_N\in\mathbb{C}\) of these matrices.
### Research Objectives
The main objective of the paper is to provide a topological expansion for the string coefficients and the connected string coefficients of the Berezin - Karpelevich integral. Specifically, the author hopes to find a topological expansion similar to that of the Itzykson - Zuber integral, but applicable to the Berezin - Karpelevich integral. This expansion form will involve the combinatorial refinement of the monotone double Hurwitz numbers.
### Main Results
The main results of the paper can be summarized as the following two theorems:
1. **Theorem 1.2**: For any \(1\leq d\leq N\) and any \(\alpha,\beta\in Y_d\), we have:
\[
I_{MN}(\alpha,\beta)=(- 1)^{\ell(\alpha)+\ell(\beta)}(MN)^{-d}\sum_{g = -\infty}^{\infty}N^{2 - 2g}\sum_{s = 0}^{2g-2+\ell(\alpha)+\ell(\beta)}\left(\frac{N}{M}\right)^s\vec{H}_g^{\bullet}(\alpha,\beta;s)
\]
and
\[
L_{MN}(\alpha,\beta)=(- 1)^{\ell(\alpha)+\ell(\beta)}(MN)^{-d}\sum_{g = 0}^{\infty}N^{2 - 2g}\sum_{s = 0}^{2g-2+\ell(\alpha)+\ell(\beta)}\left(\frac{N}{M}\right)^s\vec{H}_g(\alpha,\beta;s)
\]
where both of these series are convergent.
2. **Theorem 3.4**: For any \(\alpha,\beta\) satisfying the Young diagrams with \(d\leq N\), we have:
\[
I_{MN}(\alpha,\beta)=(- 1)^{\ell(\alpha)+\ell(\beta)}(MN)^{-d}\sum_{g = -\infty}^{\infty}N^{2 - 2g}\sum_{s = 0}^{2g-2+\ell(\alpha)+\ell(\beta)}v^s\vec{H}_g^{\bullet}(\alpha,\beta;s)
\]
where \(\vec{H}_g^{\bullet}(\alpha,\beta;s)=\vec{W}_{2g-2+\ell(\alpha)+\ell(\beta)}(\alpha)