Sparse tensor Galerkin discretization of parametric and random parabolic PDEs---analytic regularity and generalized polynomial chaos approximation (Q2870600)

In \((0,T)\times D\), where \(D\subset R^{d}\) is a bounded Lipschitz domain, the authors consider the random parabolic initial boundary value problem \(\partial u/\partial t -\nabla\cdot(a(x,\omega)\nabla u)=g(t,x)\), \(u|_{\partial D\times(0,T)}=0\), \(u|_{t=0}=h(x)\). For the considered initial boundary value problems, analyticity of the solution with respect to the parameters is shown and an a priori error analysis for \(N\)-term generalized polynomial chaos approximations in a scale of Bochner spaces is presented. The problem is reduced to a parametric family of deterministic initial boundary value problems on an infinite dimensional parameter space by Galerkin projection onto finitely supported polynomial systems in the parameter space. The authors establish the uniform stability with respect to the support of the resulting coupled parabolic systems, and analyticity of the solution with respect to the countably many parameters is shown. A regularity result of the parametric solution for both compatible as well as incompatible initial data and source terms is proved.