pystra.active_learning.subset#
Use it: User guide · Worked example · Theory
Replicated subset simulation and estimator-driven active learning.
The nested-event construction follows Au and Beck (2001), doi:10.1016/S0266-8920(01)00019-4. Gaussian-preserving pCN proposals replace componentwise modified Metropolis (Cotter et al., 2013, doi:10.1214/13-STS421). Active enrichment follows the modular framework of Moustapha, Marelli and Sudret (2022), doi:10.1016/j.strusafe.2021.102174; this is a documented variant of that workflow.
Classes
One level's threshold, conditional probability and chain diagnostics. |
|
One independent replication, including incomplete-run diagnostics. |
|
Subset simulation with replicated sampling and adaptive enrichment. |