Morris sample problem in boundary-distribution definition: bounds are not legal. #329

[MatthiVH]

Same issue as #329 but now for the Morris sampler.

The saltelli sampler can handle other distributions in the problem definition, however the Morris sampler can not.

I assume the Morris sampler doesn't recognise the 'dists'-statement in the problem-definition and assumes all uniform.
Can this be fixed?

Regards,
Matthias

[ConnectedSystems]

I'll have a look at this sometime over the next few days - finding time is an issue at the moment so I can't promise anything concrete.

[MatthiVH]

No problem, that's perfect.

[ConnectedSystems]

Hi @MatthiVH

I'm not sure when the next release will be made, but in the meantime you can try out the "bleeding edge" development version:

Please do note that this version has a number of changes and has not been extensively tested.

pip install git+https://github.com/SALib/SALib.git@experimental-methods

Specific to your case, sampling non-uniform distributions may not work with groups of parameters (for Morris in particular).