Currently, the infer() method of a ProPPA model runs a single sampling chain to infer parameters. It may be useful to have an option for multiple chains, to provide different initialisations. A subsequent decision is how to display the results: separately per chain (eg to illustrate potential issues related to bad initialisation) or jointly (for a more accurate picture obtained by more samples).)
Currently, the infer() method of a ProPPA model runs a single sampling chain to infer parameters. It may be useful to have an option for multiple chains, to provide different initialisations. A subsequent decision is how to display the results: separately per chain (eg to illustrate potential issues related to bad initialisation) or jointly (for a more accurate picture obtained by more samples).)