Solving semilinear differential equations by probabilistic methods (Q1314946)

Let \(D\) be a bounded domain in \(R^ d\) with smooth boundary \(\partial D\). The author gives a probabilistic method to prove the existence of a solution of the semilinear partial differential equation \((-L+\mu_ 1) u+f_ 1(u)=\mu_ 2\) in \(D\) with the boundary condition \(Ku+{\partial u \over \partial n} + \nu_ 1u-f_ 2 (u)=\nu_ 2\) on \(\partial D\), where \(\mu_ i\) and \(\nu_ i\) are bounded signed measures belonging to the generalized Kato's class and \(L= \sum^ d_{i,j=1} {\partial\over {\partial x_ i}}(a_{ij} {\partial \over \partial x_ j})+\sum^ d_{i=1} b_ i {\partial \over \partial x_ i}+c\). Under the hypothesis of uniform ellipticity on \(L\) with additional hypothesis of the smoothness and for the Markovian property on the coefficients, there exists the reflecting diffusion process \({\mathbf M} = (x_ t,P_ x)\) which has the generator \(L\) with boundary condition \(Ku+{\partial u \over \partial n}=0\). The author uses the characterizations of the solution by means of a martingale and a Dirichlet form to get a solution as a limit of an approximating representation given by \({\mathbf M}\).