Polynomial chaos expansions (PCE), one of the most powerful and versatile surrogate models available today

• Non-intrusive PCE facilities
• State-of-the-art sparse regression solvers, including least angle regression, subspace pursuit, Bayesian compressing sensing and much more
• Support for classical polynomials, as well as polynomials orthogonal to user-defined input distributions
• Support also fully data-driven, arbitrary PCE

Sensitivity analysis can identify the driving factors that most influence the response of a model

• Approximation- and simulation- based methods
• Advanced variance decomposition techniques (e.g. Sobol' indices)
• Support for metamodel-based sensitivity analysis
• Support for sensitivity analysis of dependent input variables

Probabilistic modelling lies at the core of uncertainty quantification. Full support for complex probabilistic models based on copula and marginal representation. Powerful data-driven inference module.

• Support for many common marginal distributions
• Support for standard copulas, as well as complex vine-copula constructions
• Fully automatic inference of both marginal distributions and copula
• Advanced sampling schemes, including latin hypercube sampling, quasi-random sequences