reminder for manual

[Andrew Gelman]

We should always lead with the fast code and then explain how it’s equivalent to the non-vectorized version. For example on pp. 61-62 of manual we have multivariate regression with hier model and we start with non-vectorized. This is confusing because if I want to imitate this model myself I have to grab different code snippets.
A

[Daniel Lee]

I just updated the next manual issue with your comment.

https://github.com/stan-dev/stan/issues/1420

On Apr 23, 2015 10:48 AM, "Andrew Gelman" <gel...@stat.columbia.edu> wrote:

We should always lead with the fast code and then explain how it’s equivalent to the non-vectorized version.  For example on pp. 61-62 of manual we have multivariate regression with hier model and we start with non-vectorized.  This is confusing because if I want to imitate this model myself I have to grab different code snippets.
A

--
You received this message because you are subscribed to the Google Groups "stan development mailing list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to stan-dev+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Bob Carpenter]

OK --- I'll fix that. There are probably a bunch more
cases of unoptimized or wrongly ordered code throughout.

I'm torn between providing program fragments, which are
easy to discuss, and trying to dump out whole programs
and then comment on them all at once. There's something
to be said for both ways of doing it.

One option is to rethink example model delivery and
break them up into a bunch of standalone examples a la
the way BUGS did it. We could do them in knitr, IPython Notebooks,
shell scripts, etc.

- Bob

> On Apr 24, 2015, at 12:48 AM, Andrew Gelman <gel...@stat.columbia.edu> wrote:
>
> We should always lead with the fast code and then explain how it’s equivalent to the non-vectorized version. For example on pp. 61-62 of manual we have multivariate regression with hier model and we start with non-vectorized. This is confusing because if I want to imitate this model myself I have to grab different code snippets.
> A
>

[Andrew Gelman]

Yes, and this will all be much less of an issue once we have that new book. So in that sense we don’t need to kill ourselves getting the examples all straight in the manual, at least not yet.