On a Toeplitz determinant identity of Borodin and Okounkov (Q1587852): Difference between revisions

Let \(T_n (\varphi)\) denote the Toeplitz matrix \((\varphi_{i-j})_{i,j=0}^{n-1}\) and \(D_n (\varphi)\) its determinant. It is supposed that \(\varphi\) has geometric mean 1 and the representation \(\varphi=\varphi_+\varphi_-\) with \(\varphi_+ (0)=\varphi_- (\infty)=1\). The matrix \(K_n\) acting in \(\ell_2(\{n,n+1,\dots \})\) defined by \[ K_n (i,j)= \sum_{k=1}^\infty \left(\varphi_-/\varphi_+\right)_{i+k}\left(\varphi_+/\varphi_-\right)_{- k-j}. \] The formula of Borodin and Okounkov is \[ D_n (\varphi)=\left(\sum_{k=1}^\infty k(\log\varphi)_k (\log\varphi)_{-k}\right) \det(I-K_n) \] expresses \(D_n (\varphi)\) in terms of the Fredholm determinant of a product of two Hankel operators. The author gives two other proofs of this formula and one of these proofs admits to extend this formula to the case of block Toeplitz determinants.