Optimizer/ surrogate accuracy and construction : RBDO

UQ with UQLab Community Q&A and How To
, , ,
1

Hi all,

Thank you for the useful uqlab software and the RBDO module. I have a few questions

  1. Please are there any Pros and Cons in the choice of global or local optimizers in surrogate assisted RBDO. For example SQP vs GA etc

  2. Appreciate any insight on how to build a surrogate for an RBDO problem for a case where some or all of the design variables (DV) are part of the hard constraints( knowing that the DV are mainly defined by their CoV and bounds ). Hard constraint comprises the environmental variables ( defined by marginals and their parameters) and some DV defined by only the CoV. My concern here is the presence of the DV in constructing the surrogate.

  3. For surrogate assisted RBDO using active learning is there a way to determine the accuracy of the built surrogate ( I mean like the LOO error )?

Appreciate shared thoughts from the forum and the team / @moustapha .

Thanks always for the support.