Necessary and sufficient conditions for validity of the weak extremum principle for second-order elliptic systems (Q1357974): Difference between revisions

The authors investigate second-order elliptic systems of \(m\) equations in a bounded domain \(G\subset\mathbb{R}^n\) \((n\geq 2)\) with ``weak mixing'' \[ \sum^n_{i,j=1} a^{(s)}_{k,ij}(x) D_iD_ju_k+ \sum^m_{k=1} \sum^n_{i=1} b^{(s)}_{k,i}(x)D_iu_k+ \sum^m_{k=1} c^{(s)}_k(x)u_k= f_s(x),\;s=1,\dots,m, \] and with ``full mixing'' \[ \sum^m_{k=1} \Biggl(\sum^n_{i,j=1} a^{(s)}_{k,ij}(x)D_iD_ju_k+ \sum^n_{i=1} b^{(s)}_{k,i}(x)D_iu_k\Biggr)+ \sum^m_{k=1} c^{(s)}_k(x)= f_s(x),\;s=1,\dots,m \] with non-degenerate coefficient matrix \(c(x)= c^{(s)}_k(x)\). It is shown that a necessary condition (as analog of dissipativity) for validity of the weak extremum principle is the non-positivity of the inverse matrix \(c^{-1}(x)\) (i.e., non-positivity of all entries of the inverse matrix). Some sufficient conditions for the validity of the above principle are formulated for systems with ``weak mixing''. In particular, the authors prove the uniqueness of classical solutions to the second and third boundary value problems for domains with Lipschitz boundary.