A note on the maximum principle for second order non-variational linear elliptic equations (Q1111758)

Consider solutions \(u\in W^{2,n}(\Omega)\cap C^ 0({\bar \Omega})\) of the Dirichlet problem of non variational elliptic equations \[ Lu=f\quad in\quad \Omega,\quad f\in L^ n(\Omega);\quad u=0\quad on\quad \partial \Omega, \] where \[ Lu=\sum a_{ij}(x)\partial^ 2u/\partial x_ i\partial x_ j,\quad \sum^{n}_{i=1}a_{ij}(x)=1, \] \[ \sum^{n}_{i,j=1}a_{ij}(x)\xi_ i\xi_ j\geq \lambda | \xi |^ 2\quad (\xi \in R^ n,\quad 0<\lambda <1/n) \] and \(\Omega\) is a bounded open subset of \(R^ n\). \textit{C. Pucci} [Ann. Mat. Pura Appl., IV. Ser. 74, 15-30 (1966; Zbl 0144.358)] proved that the ``maximum principle'' \[ (1)\quad \| u\|_{L^{\infty}(\Omega)}\leq K\| Lu\|_{L^ q(\Omega)},\quad 1\leq q\leq n \] is true for \(q=n\). The author in the present paper studies how q increases to n if \(\lambda\) tends to zero. An asymptotic bounded (as \(\lambda\) \(\to 0)\) for q is given in terms of \(\lambda\).