samueldnj / hierProdAnalysis

A simulation-estimation model for that compares the performance of hierarchical multispecies and single species surplus production stock assessment models.
0 stars 0 forks source link

Allow for phase switching by control files #13

Closed samueldnj closed 7 years ago

samueldnj commented 8 years ago

Right now, msProd fixes the lnq prior mean, which is shared. This should be allowed to vary.

Ideas

samueldnj commented 7 years ago

Just turned on estimation manually, converges more often than before now that covariance is estimated. I'm still thinking that AMBD-RE might be the way forward.

edit: ADMB-RE is a bust so far.

samueldnj commented 7 years ago

This issue just graduated to implementing phase switching in the control files. This will allow for more control over the fitting process, and potentially improve the refitting subroutine.

samueldnj commented 7 years ago

This is implemented, but MS and SS models both use the same phases for the same quantities. Might be worth having a ss and ms version of each, as in some cases it might not make sense to have parameters vary:

eg. mlnq should be able to vary in the MS model, but it's not clear that it should always vary in the SS model

As a result, this should be revisited when issue #18 is resolved, as under that ctl structure it will be easier to include more variables.

samueldnj commented 7 years ago

Updated the current control file structure to have phases for most (if not all) parameters in estimation. Closing.