The generalized Segal--Bargmann transform and special functions (Q1876542)

This is an overview article based on results of the authors, G.~Zhang, B.~Ørsted, J.~Faraut, A.~Koranyi and others. The paper is within the framework of the approach to the theory of special functions as part of the representation theory of Lie groups. Let \(\mathcal D\) be a bounded symmetric domain in \({\mathbb C}^d\), \(0\in\mathcal D\). It is a homogeneous space \(G/K\), a symmetric space. We can consider \(G\) to be simple. Assume \(\mathcal D\) to be isomorphic to a tube type domain \(T(\Omega)=\Omega+i{\mathbb R}^d\) where \(\Omega\subset{\mathbb R}^d\) is a symmetric cone. The cone \(\Omega\) is the image under the Cayley transform of a submanifold \({\mathcal D}_{\mathbb R}\subset \mathcal D\). It is a real symmetric space \(H/K\cap H\) where \(H\) is a fixed point subgroup for an involution of the group \(G\) (a ``real part'' \ of \(G\)). There is a canonical series of highest weight representations of the group \(G\) on the spaces \({\mathbf H}_\nu(\mathcal D)\). Here \(\nu\) is a real parameter running in the so-called Berezin-Wallach set. The space \({\mathbf H}_\nu(\mathcal D)\) is a Hilbert space consisting of holomorphic functions on \(\mathcal D\). The authors show how to construct the Berezin transform and the generalized Segal-Bargmann transform using the ``restriction map'' \(R_\nu:{\mathbf H}_\nu({\mathcal D})\to C^\infty({\mathcal D}_{\mathbb R})\). The map \(R_\nu\) is the restriction to \({\mathcal D}_{\mathbb R}\) with multiplication by some function. The Berezin transform is \(R_\nu R^\star_\nu\) and the generalized Segal--Bargmann transform is the unitary part of \(R^\star_\nu\), i.e. \(R^\star_\nu=U_\nu(R_\nu R^\star_\nu)^{1/2}\). Let \(\varphi_\lambda(x)\) be a spherical function on \(H/K\cap H\), \(\Delta(x)\) the determinant function on the Jordan algebra \({\mathbb R}^d\). The decomposition of \(\Delta(e-x^2)^{-\nu/2}\varphi_\lambda(x)\) in elements of a \(K\cap H\)-invariant basis \(\psi_m(x)\) where \(m=(m_1,\ldots,m_r)\), \(r\) being the rank of \(G/K\), has coefficients \(p_{\nu,m}(\lambda)\), the generalized Meixner-Pollachek polynomials. For these polynomials, recurrence relations and difference relations are written (the latter contain shifts in arguments). In the unbounded realization (\(T(\Omega)\)) the map \(R_\nu\) is simply the restriction to \(\Omega\) and the unitary part becomes the Laplace transform \({\mathcal L}_\nu\). Here the authors focus their attention on Laguerre polynomials \(L^\nu_m(x)\) and Laguerre functions \({\ell}^\nu_m(x)\) on the cone \(\Omega\) introduced by Faraut and Koranyi. They provide a generating function for the Laguerre polynomials and recurrence relations for the Laguerre functions. For the case when \(\Omega\) is the cone of positive definite Hermitian matrices, some differential equations for the Laguerre functions are given. The Laplace transform \({\mathcal L}_\nu\) transfers the Laguerre functions \({\ell}^\nu_m(x)\) to elements of an orthogonal basis in the space \({\mathbf H}_\nu(T(\Omega))^{K\cap H}\).