Fast high-dimensional integration using tensor networks

The design and application of regression-free tensor network representations for integration is presented. Tensor network methods are demonstrated to outperform Monte Carlo for test problems, and exponential convergence is shown to be achievable for a non-analytic integrand.