We propose a method for computing any Gelfand-Dickey τ function living in SegalWilson Grassmannian as the asymptotics of block Toeplitz determinant associated to a certain class of symbols W(t; z). Also truncated block Toeplitz determinants associated to the same symbols are shown to be τ function for rational reductions of KP. Connection with Riemann-Hilbert problems is investigated both from the point of view of integrable systems and block Toeplitz operator theory. Examples of applications to algebro-geometric solutions are given.
IntroductionThis paper deals with the applications of block Toeplitz determinants and their asymptotics to the study of integrable hierarchies. Asymptotics of block Toeplitz determinants and their applications to physics is a developing field of research; in recent years it has been shown how to compute some physically relevant quantities (e.g. correlation functions) studying asymptotics of some block Toeplitz determinants (see [25], [26], [27]). In particular in [25] and [26] authors, for the first time, showed effective computations for the case of block Toeplitz determinants with symbols that do not have half truncated Fourier series. This is of particular interest for us as, with our approach, we will be able to do the same for certain block Toeplitz determinants associated to algebro-geometric solutions of Gelfand Dickey hierarchies. Let us mention some theoretical results about (block) Toeplitz determinants we will use in this paper. Given a function γ(z) on the circle we denote T N (γ) the Toeplitz matrix with symbol γ given byWe use the term block Toeplitz for the case of matrix-valued symbol γ(z). In that case the entries γ (j−i) of the above matrix are n × n matrices themselves. We denoteand we use the notation T (γ) for the N × N matrix obtained letting N go to infinity. where G(γ) is a normalizing constant and the operator T (γ)T (γ −1 ) is such that its determinant is well defined as a Fredholm determinant (see section 2 for the precise statement). Once the asymptotics had been computed the next quite natural question was to find an expressions relating directly D N (γ), and not just its asymptotics, to certain Fredholm determinants. The problem was solved many years later by Borodin and Okounkov in [9] for the scalar case and generalized, in the same year, for matrix case by E. Basor and H.Widom in [10]. For matrix valued case Borodin-Okounkov formula reads(here we assume G(γ) = 1). The operator (I − K γ,N ) can be written explicitely in coordinates knowing certain Riemann-Hilbert factorizations of γ. Its Fredholm determinant is well defined (see section 2 for details). Now many proofs of Borodin-Okounkov formula are known (for instance [11] contains another proof of the same formula, see also the earlier paper [12]).In this paper we apply block-Toeplitz determinants to the computation of τ function of an (almost) arbitrary solution of Gelfand-Dickey hierarchy(L differential operator of order n, j = nk). More precisely to a given pointin the big cell of Segal-...