A regularity theory for stochastic partial differential equations with a super-linear diffusion coefficient and a spatially homogeneous colored noise (Q2021414): Difference between revisions

The authors consider an equation \[ du=(a^{ij}u_{x_ix_j}+b^iu_{x_i}+cu)\,dt+\xi h(u)\,dF \] on \((0,\tau]\times\mathbb R^d\), where \(\tau\) is a bounded positive stopping time, the random initial condition \(u_0\) is non-negative, the coefficients \(a\), \(b\), \(c\), \(\xi\) and \(h\) are predictable, \(a\), \(b\), \(c\) are uniformly bounded in \(C^2(\mathbb R^d)\) in the spatial variable, \(a\) is uniformly elliptic, \(\xi\) is uniformly bounded in some \(L_\sigma(\mathbb R^d)\) in the spatial variable for a specified \(\sigma\) and \(F\) is a centered Gaussian noise white in time and coloured in space (the spatial correlation \(f\) is assumed to be a non-negative and positively semi-definite distribution with the measure \(\hat{f}\) having a certain moment on the unit ball in \(\mathbb R^d\) if \(d\ge 2\) or \(f\) is simply a continuous function). It is assumed that either \(h\) is uniformly Lipschitz continuous and \(L^p\)-integrable in time or \(h(u)=|u|^{1+\lambda}\) for \(\lambda\ge 0\) subject to an upper bound specified in the paper and the initial condition \(u_0\) is assumed to belong to \(L^p(\Omega;H^{\gamma-2/p}_p)\), where the ranges of \(p\) and \(\gamma\) are specified. If \(h(u)=|u|^{1+\lambda}\) then \(u_0\) is further assumed to belong to \(L^1(\mathbb R^d)\). Then the equation above has a unique solution \(u\) with space-time Hölder continuous paths that are \(p\)-integrable in \(H^\gamma_p\).