CmdStan Guide now online

yes!! I’ve fixed that in the Rmd, haven’t republished.

@mitzimorris Will the online CmdStanR guide eventually have a URL that links to the latest version rather than having to specify a particular version number? I was going to link to it from the CmdStanR doc but instead of the current link with a version number

we’d need something like

to point to the latest version so that we don’t need to keep updating URLs everywhere we link to it. Does that make sense?

yes, will create the unversioned-redirects now.
just created the unversioned redirects - above link should work.

1 Like

Awesome, thanks Mitzi!

We rarely want to make inference for expectation (or quantiles) of lp__ so ESS or MCSE are not relevant in that sense, but Rhat and ESS for lp__ are additional diagnostics which may reveal reasons for inefficiencies in the mixing speed. There might something better, but I would report them at the moment.

1 Like

Reporting all stats for lp__. Also continuing to report these stats for all sampler params as well, because objections were raised (see GitHub issue), but added a blank line between sampler parameter summaries and model parameter summaries.

The new stansummary program has better command-line option handling, allows for your choice of quantiles to display, and the output is a little more readable - see description here:

Many thanks to everyone who’s contributed to this thread for making the docs so much better!

The 2_24 versions are up. If you want to link to these docs, use the unversioned URLs: