Consistency and stability of a Milstein-Galerkin finite element scheme for semilinear SPDE (Q487684)

The author is interested in approximating the mild solution \(X:[0,T]\times \Omega\rightarrow H\) to the semilinear SPDE \[ dX(t)+[AX(t)+f(X(t))] dt = g(X(t))dW(t), \qquad 0\leq t\leq T, \] with \(X(0)=X_0\). Here, \(H\) is a separable Hilbert space, \(-A:\mathrm{dom}(A)\subset H \rightarrow H\), \(W\) a cylindrical \(Q\)-Wiener process, \(f\) and \(g\) denote Lipschitz continuous functions. To this purpose, he introduces a Milstein-Galerkin finite element scheme and proves that this scheme satisfies the properties of a class of abstract numerical one-step schemes for which he develops a stability and consistency analysis. The notions of residual operator and bistability, already used by the author in previous papers, turn out quite useful. In the setting of a semilinear SPDE, the residual operator is called the local truncation error. It is measured in terms of the stochastic Spijker norm and is used to estimate the strong error from above and below. The result provides a slightly sharper order of convergence than in previous works. The paper ends with an analysis of noise approximation and truncation.