Which document do you mean? The R package examples and vignettes go into quite a bit of detail regarding building a Bayesian model fit, including priors etc. This is discussed in the paper at some length too.

I meant in libprofit and readthedocs. I did not install the R nor the python package yet. I just downloaded and compiled libprofit, to start getting a feeling with the code by using the cli

ok, now I see on p.1534 of the paper that libprofit does not include a built-in optimization, sorry about that, I will install the python interface.

The R interface is quite a bit more advanced in terms of features, but people are obviously encouraged to expand the Python side as suits them too. A whole bunch of user vignettes are now kept online too: http://rpubs.com/asgr/

is it possible to have a todo list of missing features in python that are implemented in R? I can read R, but still that would help (alternatively, build a series of issues related to fetaures to be implemented)

The main difference is the R side has a bunch of helper functions that create inputs for ProFit- e.g. profitMakeSigma for building sigma maps, profitProFound for making segmentation maps and extracting good initial parameter estimates. These are the same type of inputs that GalFit needs, so if you have strategies to prepare inputs for GalFit then you might not need these helper functions. The advantage for us (and why we made them) is that it keeps the automation / scripting self-contained within R, which has a lot of benefits in terms of software dependencies and simplification. I'm aware some people have forked the Python variant and adapted it, but more in order to interface with different optimisers (e.g. DNEST). I will have a chat with Rodrigo and think about putting together a hit-list of functions and features. Alas, I don't use Python at all (can read it, but I've never written any).