Every P-convex subset of R^2 is already strongly P-convex

From MaRDI portal
Publication:658324



Abstract: A classical result of Malgrange says that for a polynomial P and an open subset Omega of Rd the differential operator P(D) is surjective on Cinfty(Omega) if and only if Omega is P-convex. H"ormander showed that P(D) is surjective as an operator on mathscrD′(Omega) if and only if Omega is strongly P-convex. It is well known that the natural question whether these two notions coincide has to be answered in the negative in general. However, Tr`eves conjectured that in the case of d=2 P-convexity and strong P-convexity are equivalent. A proof of this conjecture is given in this note.


Let \(P(D)\) be a partial differential operator with constant coefficients in \(n\) variables and let \(\Omega\subset \mathbb{R}^n\). It is well known that surjectivity of \(P(D)\) on \(C^\infty(\Omega)\) (i.e. \(P\)-convexity of \(\Omega\)) does not imply surjectivity of \(P(D)\) on the distributions \(D'(\Omega)\) (i.e. strong \(P\)-convexity of \(\Omega\)) in general while the inverse implication is always true. By a conjecture of Treves both convexity conditions should be equivalent if \(n=2\). This longstanding conjecture is proved in the present paper. The proof is based on a careful study of the extension of regularity for solutions of \(P(D)\) in cones using Hörmander's theory based on the function \(\sigma_P\) which can be calculated using the localizations of \(P\) at infinity. Since the principle part \(P_m\) of a polynomial \(P\) in two variables has only finitely many zeros on the real unit sphere, one can show that each vector \(y\in \mathbb{R}^2\) with \(\sigma(y)=0\) is characteristic if \(n=2\). Using Hörmander's geometric description of \(P\)-convexity and its analogue for strong \(P\)-convexity the proof of Treves' conjecture is obtained.











This page was built for publication: Every \(P\)-convex subset of \({\mathbb{R}^2}\) is already strongly \(P\)-convex

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q658324)