Local \(L_\infty\)-estimates, weak Harnack inequality, and stochastic continuity of solutions of SPDEs (Q338463): Difference between revisions

A parabolic stochastic partial differential equation \[ du_t=L_tu_t dt+M^k_t u_t dw^k_t \] for \((t,x)\) in \([0,4]\times\{z\in\mathbb R^d:|z|<2\}\) is considered, where \(L\varphi=\partial_i(a^{ij}\partial_j\varphi)\), \(M^k\varphi=\sigma^{ik}\partial_i\varphi\), \(w=(w^k)\) are independent Wiener processes and \(a\) and \(\sigma\) are bounded, predictable and satisfy \(2a+\sigma\sigma^*\geq\lambda I\) for some \(\lambda>0\). The authors first prove estimates of the form \[ P\,(\|f(u)\|^2_{L^\infty(I_1\times A_1)}\geq C\kappa\alpha,\,\|f(u)\|^2_{L^2(I_2\times A_2)}\leq\alpha)\leq c\kappa^{-\delta}, \] given a suitable function \(f\), for every \(\alpha>0\) and \(\kappa\geq 1\), i.e., they estimate the \(L^\infty\)-norm over a smaller space-time set \(I_1\times A_1\) by the \(L^2\)-norm over a larger space-time set \(I_2\times A_2\). Secondly, they prove a stochastic lower estimate for non-negative solutions \(u\), and finally, they prove that the solutions are almost surely continuous at every given inner point \((t_0,x_0)\) in \((0,4)\times\{z\in\mathbb R^d:|z|<2\}\).