search

yabebalFantaye / MCEvidence

A python package implementing the MARGINAL LIKELIHOODS FROM MONTE CARLO MARKOV CHAINS algorithm described in Heavens et. al. (2017)
MIT License
38 stars 17 forks source link

Choosing parameters on MCEvidence #11

Open kssiobr opened 5 years ago

kssiobr commented 5 years ago

Hi, I tried to run the example " MLE = MCEvidence('/path/to/chain').evidence() " para I got errors cause some of the derived parameters return negative or complex eingvalues.

If I use: MLE = MCEvidence('/path/to/chain', ndim = 33).evidence() am I sure the code is taking the first 33th parameters, right?

And how do I say that I want the CROSS evidence method? Something like MLE = MCEvidence('/path/to/chain', ndim = 33, cross=True).evidence() ??

Is there any way MCEvidence uses the chains loaded from GetDist via loadMCSamples?

ps: Does it make any sense to use derived parameters to evaluate evidence anyway?

Thanks a lot!

ivandebono commented 5 years ago

As the number of dimensions increases, the method becomes less reliable, and the computation takes much longer. With 33 parameters, I expect your computation will take hours instead of minutes.

The short answer to your quesiton about derived parameters is that they should not be included.