Dear @ali,
Thanks for the comments. Similar to conducting UQ for a deterministic model, several aspects can be developed for analyzing a stochastic simulator, namely uncertainty propagation, sensitivity analysis, and reliability analysis. Generally speaking, Monte Carlo simulations can be applied for these analyses. Hence, UQLab can be still used as far as Monte Carlo methods are concerned (e.g. for sensitivity analysis, you can read our recent paper[1] for details).
As you mentioned, surrogate models can typically help us alleviate the computational burden. That is why we launched the SAMOS project. However, the metamodels[2][3] that we are developing are not available in UQLab yet (but will certainly be included in a future version). Nevertheless, if you want to perform a specific analysis for a case study in structural engineering, we can have a more detailed discussion.
Best regards,
Xujia
-
X. Zhu and B. Sudret, “Global sensitivity analysis for stochastic simulators based on generalized lambda surrogate models”, 2020. arXiv:2005.01309 ↩︎
-
X. Zhu and B. Sudret, “Replication-based emulation of the response distribution of stochastic simulators using generalized lambda distributions”, Int. J. Uncertainty Quantification, vol. 10, pp. 249–275, 2020. DOI:10.1615/Int.J.UncertaintyQuantification.2020033029 ↩︎
-
X. Zhu and B. Sudret, “Emulation of stochastic simulators using generalized lambda models”, 2020. arXiv:2007.00996 ↩︎