Algebraic aspects of the Dirichlet problem

From MaRDI portal





The very interesting paper under review deals with some algebraic aspects of the classical Dirichlet problem for the Laplace equation in the unit \(n\)-dimensional ball \(\mathbb B^n:\) \[ \begin{cases} \Delta u=0 & \text{in}\;\mathbb B^n,\\ u=f & \text{on}\;\mathbb S^{n-1}=\partial\mathbb B^n \end{cases} \tag{1} \] with continuous data \(f\) on the sphere \(\mathbb S^{n-1}.\) As well known, the solution \(u\) to \((1)\) is given by the Poisson integral \[ u(x)=\int_{\mathbb S^{n-1}} f(y)K_n(x,y)dS_y \] where \(K_n(x,y)\) is the Poisson kernel \[ K_n(x,y)=\omega_n \frac{1-| x| ^2}{| x-y| ^2} \] and \(\omega_n\) is reciprocal to the surface area of \(\mathbb S^{n-1}.\) Although many qualitative properties of the solution \(u,\) such as regularity up to the boundary, analyticity etc., follow from that formula, authors address some interesting questions of algebraic nature which are not easy outgrowths of Poisson's formula. It is a classical result that, if \(f(x)\) is a polynomial of degree \(m,\) then the solution \(u(x)\) of problem \((1)\) is a polynomial of degree at most \(m.\) The authors take this fact as a starting point in order to study wider classes of algebraic functions preserved by the Poisson operator. More precisely, the following two questions are considered: Question A. Assume the data \(f\) is a \textit{rational} function without poles on \(\mathbb S^{n-1}.\) Is then the solution \(u(x)\) of the Dirichlet problem \((1)\) necessarily rational? Question B. Assume the data \(f\) is an \textit{algebraic} function which is analytic in a neighborhood of \(\mathbb S^{n-1}\) in \(\mathbb R^n.\) Is then the solution \(u(x)\) of the Dirichlet problem \((1)\) necessarily algebraic? Somewhat surprisingly, even though the answer to Question A is affirmative for \(n=2\) (as it follows from a more general result of the first author), it is negative for all odd \(n\geq 3\) and, at least, for all even \(n\) with \(4\leq n\leq 270.\) The answer to Question B is negative in all dimensions. The main results of the paper provide examples of rational functions \(f\) in \(\mathbb R^n\) for all odd \(n\geq 3\) and even \(n\geq 4\) with an additional algebraic condition satisfied, which yield non-rational -- even non-algebraic -- solutions to the Dirichlet problem \((1).\) In case \(f\) belongs to certain subclasses of rational functions in \(\mathbb R^n\) with \(n\) even, some positive results on Question A are given. Finally, examples of algebraic data \(f\) in \(\mathbb R^2\) are proposed for which the solutions are non-algebraic. The machinery employed in obtaining all these interesting results relies on Laurent decomposition, ultraspherical polynomials and the Nehari transform. For the entire collection see [Zbl 1062.30001].











This page was built for publication: Algebraic aspects of the Dirichlet problem

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5703501)