Matzoh ball soup: A symmetry result for the heat equation (Q923231)

The starting point of the paper is a question mentioned by \textit{L. Zalcman} [Contemp. Math. 63, 337-350 (1987; Zbl 0641.31003)] viz. ``Let \(\Omega\) be a bounded solid. Let U be the solution of the following initial-boundary value problem for the heat equation \[ (Ia)\quad (\partial /\partial t)U-\Delta U=0\text{ in } \Omega \times (0,\infty); \] \[ (Ib)\quad U=1\text{ on } \partial \Omega \times (0,\infty);\quad (Ic)\quad U=0\text{ on } \Omega \times \{0\}. \] Suppose that the level surfaces of U(.,t) are invariant with respect to the time variable \(t>0\). Is \(\Omega\) a ball?'' The author derives the following results: A. Let \(\Omega\) be a bounded connected open set in \({\mathbb{R}}^ n\), having \(C^ 2\)-smooth boundary \(\partial \Omega\), and let U be the solution of problem I above. Then \(\Omega\) is a ball if (i) for every \(\tau >0\) and for every level surface \(\Gamma\) (whether connected or not) of U(.,z), \(U|_{\Gamma \times [\tau,\infty)}\) depends only on t, or if (ii) for every \(\tau >0\) and for every connected component \(\Sigma\) of every level surface of U(.,\(\tau\)), \(U|_{\Sigma \times [\tau,\infty)}\) depends only on t, and \(\Omega\) is a solid. B. Let \(\Omega\) be as in A above. Let g,g\(\neq 0\), be any function in \(L^ 2(\Omega)\) and let U satisfy Ia, Ib and the initial condition \(U=1- g\) on \(\Omega\times \{0\}.\) If U satisfies condition (i) of A above the one of the following two cases occurs: (i) g is an eigenfunction of the Laplacian in \(\Omega\) with zero Dirichlet conditions, (ii) \(\Omega\) is a ball, and, for every \(t>0\), g and U(.,t) have constant sign and are radially symmetric with respect to the centre of \(\Omega\).