On the analyticity of solutions to semilinear differential equations degenerated on a submanifold (Q712176)

The authors consider the semilinear equation \(Pu= F[u]\). The linear part \(P\) is a partial differential operator in \(\mathbb{R}^N\), \(N= n+ m\): \[ P= \sum_{\alpha,\beta,\gamma\in A} c_{\alpha,\beta,\gamma} x^\gamma D^\alpha_x D^\beta_y,\quad x\in\mathbb{R}^n,\quad y\in \mathbb{R}^m, \] where the set of indices \(A\) and the constants \(c_{\alpha,\beta,\gamma}\in \mathbb{C}\) satisfy the conditions by \textit{V. V. Grushin} in [Math. USSR, Sb. 13, 155--185 (1971; Zbl 0238.47038)], granting hypoellipticity and analytic hypoellipticity for the corresponding linear equations. The semilinear part is of the form \[ F[u]= G(x,y,x^\gamma D^\alpha_x D^\beta_y u)_{\alpha,\beta,\gamma\in B}, \] where \(G\) is analytic function of its arguments and the multi-indices in \(B\) are lower order with respect to those in \(A\), in a suitable sense. The authors prove that, under an initial Sobolev-type regularity for \(u\), the solutions of the semilinear equation are analytic. The remarkable proof is based on the construction of a fundamental solution for \(P\).