Uniqueness of the Cauchy problem for a class of quasihomogeneous operators. (Q1066004)

The author considers in an open subset \(\Omega\) in \(R^{n+m}\) a second order operator of the form \[ P=\sum_{| \alpha | =2}a_{\alpha}(x,y)D_ x^{\alpha}+p_ 1(x,y,D_ y)+ip_ 2(x,y,D_ y\quad), \] where \(x\in R^ n\), \(y\in R^ m\), the principal symbol of P is real and has coefficients in \(C^{\infty}(\Omega)\), and for \(j=1,2\), \(p_ j(x,y,D_ y)\) is a homogeneous operator of order 1 with real symbol with coefficients in \(C^{\infty}(\Omega)\). The author is interested in the role of the lower order terms in the uniqueness of the Cauchy problem, and to this end he proves two theorems. In the first theorem it is assumed that \(\Omega\) is a neighborhood of a point \((x_ 0,y_ 0)\) at which one of the following is true: (1) the linear forms \(p_ 1(x_ 0,y_ 0,\eta)\) and \(p_ 2(x_ 0,y_ 0,\eta)\) are independent, or (2) \(p_ 2\equiv 0\) on \(\Omega\) and \(p_ 1(x_ 0,y_ 0,\eta)\not\equiv 0\). It is also assumed that \(S=\{(x,y)\in \Omega:\) \(\phi (x,y)=\phi (x_ 0,y_ 0)\}\) is a non-characteristic oriented hypersurface which is pseudo-convex with respect to the bicharacteristics of P at \((x_ 0,y_ 0)\). Under certain additional conditions on \(p_ 1,p_ 2\), it follows that there exists a neighborhood V of \((x_ 0,y_ 0)\) such that if \(u\in C^{\infty}(\Omega),| Pu(x,y)| \leq C\{| \nabla_ xu(x,y)| +| u(x,y)| \}\) and \(u=0\) for \(\phi (x,y)<\phi (x_ 0,y_ 0)\), then \(u\equiv 0\) in V. This theorem is related to the work of \textit{R. Lascar} and \textit{C. Zuily} [Duke Math. J. 49, 137-162 (1982; Zbl 0536.35001)]. The second theorem concerns the operator \[ P=D^ 2_ t+\sum^{r}_{i\leq j=1}a_{ij}(t,x,y)D_ iD_ j+p_ 1(t,x,y,D_ y)+ip\quad_ 2(t,x,y,D_ y) \] where \(1\leq r\leq n-1\), \(t\in R\), \(x\in R^{n-1}\), \(y\in R^ m\), \(D_ i\) stands for differentiation with respect to \(x_ i\), and S is a non-characteristic hypersurface defined in the neighborhood of the origin by \(t=0\). Here, S is not assumed pseudo-convex, but under certain conditions on \(p_ 1,p_ 2\), there still follows a conclusion analogous to that of the first theorem. This second theorem is related to the work of \textit{S. Alinhac} and \textit{C. Zuily} [Commun. Partial Differ. Equations 6, 799-828 (1981; Zbl 0482.35052)]. The proofs of the theorems utilize the technique of Carleman inequalities.