Bayesian inference with multiple forward model with different numbers of provided measurements but with one common discrepancy model and parameter?

UQ with UQLab Community Q&A and How To
7

Dear Paul-Remo,

thank your for your answer, especially the pre-version of the new part in the documentation and new versions of the module files. I will start to test, after informing you about a suggestion for a fix for a well-hidden bug that is also active (and hidden) in your new version of uq_initialize_uq_inversion.m

I suggest to replace the in the lines 89-90 in uq_initialize_uq_inversion.m the name of the variable “ForwardModel” by an other one, for example “ForwardModelFromOptValue” to avoid that in line 415 the variable “ForwardModel” on the left-hand side is accidentally already defined as up_model.
This has produced some quite strange error message when I used
BayesOpts.ForwardModel = myForwardModel with myForwardModel being a single UQLab-Model instead of relying, as most other users do, on the default behavior of UQ-Lab to use the last created MODEL object (in this case also myForwardModel) as the forward model if uq_creasteAnalysis is called.

When I evaluated SCR_BI_to_233_test.m (1.5 KB) I got the
error message:

Unrecognized property ‘Model’ for class ‘uq_model’.

Error in uq_initialize_uq_inversion (line 415)
ForwardModel(ii).Model = uq_createModel(ModelOpt,‘-private’);

Error in SCR_BI_to_233_test (line 48)
myBayesianAnalysis = uq_createAnalysis(BayesOpts);

And I must admit that I need some debug runs to realize that the problem was not
in the evaluation of uq_createModel(ModelOpt,’-private’).

Best regards
Olaf

1 Like