1Blurb::
2Calculate model evidence using a Monte Carlo sampling approach
3
4Description::
5The \c mc_approx keyword for model evidence indicates that
6sample values will be generated from the prior distribution,
7and then the simulation model will be evaluated at these
8sample values to obtain corresponding likelihood values.
9The average of the likelihood weighted by the prior is the
10model evidence. The accuracy of this approximation depends
11on the number of samples taken, which is specified by
12the \c evidence_samples keyword.
13
14<b> Default Behavior </b>
15
16If \c evidence_samples is not specified with \c mc_approx, Dakota
17uses the number of chain samples from the MCMC (\c chain_samples)
18as the number of samples to use for calculating the model evidence.
19
20<b> Expected Output </b>
21Currently, the model evidence will be printed in the screen output
22with prefacing text indicating if it is calculated by
23Monte Carlo sampling.
24
25<b> Usage Tips </b>
26
27
28Topics::
29
30Examples::
31
32Theory::
33Faq::
34See_Also::
35