Explore PHYGRAV surrogate applications. Each workspace includes inputs, sample scenarios, and outputs tailored to partners.
Direct access to published surrogate apps.
Surrogates approximate tsunami simulations with generalized Polynomial Chaos Expansions (gPCE) built from a sparse set of high-fidelity runs.
Stochastic inputs (e.g., slide volume, location) are modeled with independent priors; uniform ranges follow maximum-entropy when data are scarce.
Hazard quantities ξ are expanded on Legendre polynomial bases: ξ̂(R)=Σ_{|α|≤p} ξα Ψα(R).
Deterministic coefficients come from sparse-grid quadrature (PSA/Smolyak), limiting full simulations while keeping accuracy.
Surrogates deliver moments, sensitivities, and rapid scenario sweeps; validation compares gPCE predictions to withheld simulations.