Data-Driven Sparse Polynomial Chaos Expansion for Models with Dependent Inputs

Polynomial chaos expansions (PCEs) have been used in many real-world engineering applications to quantify how the uncertainty of an output is propagated from inputs. PCEs for models with independent inputs extensively explored literature. Recently, different approaches proposed dependent expand use more applications. Typical include building based on Gram-Schmidt algorithm or transforming into However, two their limitations regarding computational efficiency and additional assumptions about input distributions, respectively. In this paper, we propose a data-driven approach build sparse The recursively constructs orthonormal polynomials using set monomials correlations output. not only reduces number minimally required observations but also improves numerical stability efficiency. Four examples are implemented validate algorithm.