Numerical solutions of Burgers' equation with random initial conditions using the Wiener chaos expansion and the Lax-Wendroff scheme (Q2455435)

The author considers the following Burgers' equation with random initial condition: \(v_t+ f(v)_x=0\), \(v(x,0,\xi)=g(x,\xi)\) with \(f(x)=\frac{1}{2}\). The argument \(\xi\) in \(g(.,.)\) is a standard Gaussian random variable. Using the Wiener chaos expansion, the random solution of the above equation is represented as \(v(x,t,\xi)=\sum_{n=0}^{\infty}\frac{1}{\sqrt{n}}v^n(x,t)\xi_n(\xi)\), where \(v^n(x,t)\) and \(\xi_n(\xi)\) are the Hermite-Fourier coefficients and Wick polynomials of order \(n\), respectively. Applying this representation to the solution in the above equation, the author develops a modification of the Lax-Wendroff scheme for the deterministic conservation law. The resulting method is tested numerically concerning stability, accuracy and efficiency. Computed are the first moments of the solution and a comparison is made with a reference solution obtained by using a Monte Carlo method.