Active-learning reliability

Compose surrogates, learning functions, reliability estimators and stopping criteria. See the Active-learning methods and literature guide for method selection and assumptions.

Public entry points

Object	Purpose
ActiveLearning	Compose and run a surrogate-assisted analysis.
KrigingSurrogate	Fit a Gaussian-process response model.
PCESurrogate	Fit dense or adaptive sparse Hermite polynomials.
PCKrigingSurrogate	Fit a polynomial trend with a Kriging residual.
MonteCarloEstimator	Estimate probability on a separate sample.
ImportanceSamplingEstimator	Use explicit Gaussian-mixture proposal centers.
SubsetSimulationEstimator	Use replicated conditional sampling.
BetaBounds	Stop on a surrogate beta-band criterion.

Use it: Active-learning methods and literature · Active learning for reliability analysis · Active-learning reliability

Module details

pystra.active_learning

Composable active-learning methods for structural reliability.