Do the generalized polynomial chaos and Frobenius methods retain the statistical moments of random differential equations? (Q1941227)

The generalized polynomial chaos method (gPC) and the random Frobenius method for solving random differential equations are compared by examining the numerical results obtained by these two methods when the authors are applied to approximate the solution of \(\ddot x(t)+ t\xi x(t)= 0\) with deterministic initial conditions. In the numerical experiments first \(\xi\) is chosen to be a uniform random variable on the interval \([0, 1]\) and then the experiments are repeated with \(\xi\) chosen to be a normal random variable with mean \({1\over 2}\) and variance \({1\over 12}\). It is found that the approximations of the first three statistical moments generated by the two methods agree closely with each other when the appropriate orthogonal polynomial basis is used in the gPC method.