UQWorld

Infer Marginals and Copula

Dear UQLab

Thanks to UQTeam, it’s possible to infer marginals and copula from given a multivariate problem.
I have 17 independent random variables (RV) and want to infer both marginals and copula from them.
Unfortunately, I faced the following warning/error from UQLab:

==============================================================

Requested input range is too large, please select a smaller range.

==============================================================

I was wondering if there is any solution to my problem?

In my given multivariate problem, the RVs denote to different aspects of my problem and are classified into unites of meter, MPa and so on.
I read the block independence example along with all other related examples. Maybe I must split and categorize my RVs into different inter-independent blocks?
Is there any guideline to infer marginal and copula from high dimensional problems in UQLab?
I really appreciate your constructive comments and any suggestions.

Best regards
Ali

Hi Ali,

on the error message you get: did you specify a family for your marginal or copula distribution? For instance, a uniform marginal distribution? Some distribution parameters are limited within certain ranges to satisfy consistency checks.

On your block independence question: uqlab provides the possibility to test for block independence automatically, sparing you the trouble to do it yourself. This is actually the default option if you specify that the copula must be inferred and don’t provide other information on it. The inference manual provides one or two examples. There are also dedicated examples ready to run in the UQlab software (have a look at the input module).

Hope this helps.
Best,
Emiliano

2 Likes

Dear @torree

First and foremost, I really appreciate the sustainable supports of UQTeam.
I read all related manuals and examples from RSUQ.
About your question, I assigned proper marginals and independent copula to my RVs.
The file is attached for further investigation.
myCopulaGenerator.m (4.2 KB)

Best regards
Ali