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
9

Dear Paul-Remo,

I have again used UQLab in a way not planed by the developers, such that I found some further problems that I have now formulated within the feature request uq_postProcessInversion: Extension of docmentation to deal with multiple calls of this function, code corrections and extension .

I would like to point out that using your new files I get from you , uq_Example_Inversion_04_PredPrey.m generates the following plot for the second mode:

uq_Example_Inversion_04_PredPrey-Lynx-with-prio
uq_Example_Inversion_04_PredPrey-Lynx-with-prio798×600 34.1 KB

and I wanted to get rid of the samples for the priori predictive distribution to
get a plot like this
uq_Example_Inversion_04_PredPrey-Lynx-without-prio
uq_Example_Inversion_04_PredPrey-Lynx-without-prio798×600 39.5 KB
.

Using your version of uq_postProcessInversion.m it holds that the call

uq_postProcessInversion(myBayesianAnalysis,…
‘badChains’, badChainsIndex,…
‘priorPredictive’, 0, …
‘posteriorPredictive’, 1000);

generated the error

Index exceeds matrix dimensions.

Error in uq_postProcessInversion (line 447)
   priorRunsCurr(:,MIndex) = model(mm).priorRuns(:,OMapCurr);

i.e. only the line number is changing with respect to the normal version of
uq_postProcessInversion.

Thanks again for your files.

Greetings
Olaf