It would be nice to be able to run further iterations of the posterior approximation, eg in case convergence is not achieved. This could be done by saving a "minimal" set of useful variables, although possibly the entire starwave instance might need to be saved.
A snippet of code in the fit_cmd method would be needed to allow restarting from the last saved posterior, instead of from the prior.
Saving every X iterations would also be desirable, in order to e.g. save progress for long runs, should they crash/terminate for some reason.
Finally, saving the entire class would e.g. allow producing useful diagnostic plots in a easy manner (eg plot several CMDs drawn from the posterior)
It would be nice to be able to run further iterations of the posterior approximation, eg in case convergence is not achieved. This could be done by saving a "minimal" set of useful variables, although possibly the entire starwave instance might need to be saved.
A snippet of code in the fit_cmd method would be needed to allow restarting from the last saved posterior, instead of from the prior.
Saving every X iterations would also be desirable, in order to e.g. save progress for long runs, should they crash/terminate for some reason.
Finally, saving the entire class would e.g. allow producing useful diagnostic plots in a easy manner (eg plot several CMDs drawn from the posterior)