A Filon-Clenshaw-Curtis-Smolyak rule for multi-dimensional oscillatory integrals with application to a UQ problem for the Helmholtz equation
Zhizhang Wu,Ivan Graham,Dingjiong Ma,Zhiwen Zhang
DOI: https://doi.org/10.1090/mcom/4007
IF: 2.118
2024-08-18
Mathematics of Computation
Abstract:In this paper, we combine the Smolyak technique for multi-dimensional interpolation with the Filon-Clenshaw-Curtis (FCC) rule for one-dimensional oscillatory integration, to obtain a new Filon-Clenshaw-Curtis-Smolyak (FCCS) rule for oscillatory integrals with linear phase over the d − d- dimensional cube [ − 1 , 1 ] d [-1,1]^d . By combining stability and convergence estimates for the FCC rule with error estimates for the Smolyak interpolation operator, we obtain an error estimate for the FCCS rule, consisting of the product of a Smolyak-type error estimate multiplied by a term that decreases with O ( k − d ~ ) \mathcal {O}(k^{-\widetilde {d}}) , where k k is the wavenumber and d ~ \widetilde {d} is the number of oscillatory dimensions. If all dimensions are oscillatory, a higher negative power of k k appears in the estimate. As an application, we consider the forward problem of uncertainty quantification (UQ) for a one-space-dimensional Helmholtz problem with wavenumber k k and a random heterogeneous refractive index, depending in an affine way on d d i.i.d. uniform random variables. After applying a classical hybrid numerical-asymptotic approximation, expectations of functionals of the solution of this problem can be formulated as a sum of oscillatory integrals over [ − 1 , 1 ] d [-1,1]^d , which we compute using the FCCS rule. We give numerical results for the FCCS rule which illustrate its theoretical properties and show that the accuracy of the UQ algorithm improves when both k k and the order of the FCCS rule increase. We also give results for both the quadrature and UQ problems when the underlying FCCS rule uses a dimension-adaptive Smolyak interpolation. These show increasing accuracy for the UQ problem as both the adaptive tolerance decreases and k k increases, requiring very modest increase in work as the stochastic dimension increases, for a case when the affine expansion in random variables decays quickly.
mathematics, applied