A two-space dimensional semilinear heat equation perturbed by (Gaussian) white noise (Q5956494): Difference between revisions

The authors consider semilinear stochastic partial differential equations of parabolic type which are perturbed by a space-time Gaussian white noise with space-dimension \(d \geq 2\) that have a form \[ \frac{\partial U}{\partial t}=LU+{\lambda}:f (AU)\: {\chi}+{\dot{W}}, \quad U(0,\cdot)=U_0, \] where \(\chi\) is a smoothed indicator function, \(L\) is a symmetric uniformly elliptic operator with bounded real analytic coefficients, \(f\) is the Fourier transform of a complex measure with compact support, \(\:\;\:\) means Wick (re)ordering with respect to the free solution, constants \(\lambda, A\) are small parameters, the sample paths of white noise \({\dot{W}}\) are tempered distributions. How to understand the equation itself and its solution? It is possible to consider the equation on any test function or to reduce it with the help of perturbed semigroup \(P\) that has \(L\) as a generator. It is established that for \(d \geq 2\) these approaches are equivalent in the class of random Colombeau-Oberguggenberger generalized functions. Existence and uniqueness of the solution in the class mentioned above is established and an explicit construction of the solution is given. For small \(A\) the authors obtain an asymptotic expansion in powers of \(\lambda\) of the solution and show that the first term of the expansion is the free solution and all the others are classical square integrable processes which are explicitly given linear combinations of iterated integrals involving Wick-reordered functions of the free solutions. A free solution, in the authors' definition, is obtained under \(\lambda=0\).