Space semi-discretisations for a stochastic wave equation (Q2498494)

The authors consider the strong approximation of the solution of the following stochastic wave equation \[ \frac{\partial^2 u}{\partial t^2}(t,x) = \frac{\partial^2 u}{\partial x^2}(t,x) + f(t,x,u(t,x))+ \sigma (t,x,u(t,x))\frac{\partial^2 W}{\partial t \partial x }(t,x) \] with initial condition \(u(0,x)=u_0\), \(\frac{\partial u}{\partial t}(0,x)=\nu_0\), \(x\in (0,1)\) and Dirichlet boundary conditions \(u(t,0)=u(t,1)=0\), \(t>0\), where \(W\) is a Brownian sheet on \(\mathbb{R}_+\times [0,1]\). First results on properties of the solution are provided, such as a moment bound and Hölder continuity of the trajectories of the solution. To define a semi-discrete numerical scheme the authors then rewrite the second-order in time stochastic partial differential equation into a system of two first-order in time equations and replace the second-order space derivative by the appropriate central difference and the derivative in space of the Wiener sheet by a first-order forward difference. The resulting system is reformulated in the form of a stochastic evolution equation and the strong convergence of its solution to the solution of the equation above is analysed. It is found that the rate of convergence is considerably less than the order of Hölder continuity of the trajectories. For example, in the special case of vanishing initial conditions, the sample paths are Hölder continuous in \((t,x)\) of order \(\alpha < \frac{1}{2}\), but the rate of convergence is of order \(\rho<\frac{1}{3}\). This result is in marked contrast to the results for parabolic stochastic partial differential equations.