Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I been told that there exists an integration formula, which states (or something of this sort) $$ \int_{U(N)} dU \det[(\mathbb I+XUYU^{-1})^{-r}\propto \frac{\det(1+x_iy_j)^{N-r-1}]_{i,j}}{\Delta_N(x)\Delta_N(y)}, $$ where $dU$ is the Haar-measure, $X=\text{diag}(x_1,\ldots,x_N)$ is a positive definite diagonal matrix with $x_i\neq x_j$ for $i\neq j$ (similar for $Y$), and $$ \Delta_N(x)=\prod_{1\leq i<j\leq N}(x_i-x_j) $$ is a Vandermonde determinant.

This is somewhat similar to a Harish-Chandra-Itzykson-Zuber integral, but I haven't seen this specific kind before. Are anyone familiar with this kind of integrals, and how to solve them? I would be particular interested in references to the literature.

share|improve this question
add comment

2 Answers

up vote 0 down vote accepted

Since I found a reference which answers this question (and noone has provided any other solution) I will give a short answer to my own question.

This type of integrals has been studied in the context of two-matrix models. The integral in question has been discussed in arXiv:0804.0873 (section 3), while a larger class of integrals has been discussed in arXiv:0512056 [math-ph] (appendix A).

share|improve this answer
add comment

This statement does not seem to be correct, because if X tends to the zero matrix, the left hand side is well defined but the right hand side is not (notice that HCIZ integral does not suffer from this).

share|improve this answer
    
Ups! I had misunderstood the right hand side. I read it as if the numerator had the determinant of the identity plus something... Sorry –  Marcel Mar 24 '13 at 23:11
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.