Solving pseudo-differential equations

Abstract

In 1957, Hans Lewy constructed a counterexample showing that very simple and natural differential equations can fail to have local solutions. A geometric interpretation and a generalization of this counterexample were given in 1960 by L.H\"ormander. In the early seventies, L.Nirenberg and F.Treves proposed a geometric condition on the principal symbol, the so-called condition $(\psi)$, and provided strong arguments suggesting that it should be equivalent to local solvability. The necessity of condition $(\psi)$ for solvability of pseudo-differential equations was proved by L.H\"ormander in 1981. In 1994, it was proved by N.L. that condition $(\psi)$ does not imply solvability with loss of one derivative for pseudo-differential equations, contradicting repeated claims by several authors. However in 1996, N.Dencker proved that these counterexamples were indeed solvable, but with a loss of two derivatives. We shall explore the structure of this phenomenon from both sides: on the one hand, there are first-order pseudo-differential equations satisfying condition $(\psi)$ such that no $L^2_{\text{loc}}$ solution can be found with some source in $L^2_{\text{loc}}$. On the other hand, we shall see that, for these examples, there exists a solution in the Sobolev space $H^{-1}_{\text{loc}}$.