Vectorizing unary functions

[Daniel Lee]

Hi Rayleigh,

Mind vectorizing one function first before trickling a whole slew of issues on missing implementations?

Daniel

[Rayleigh L]

Hi Daniel,

I have vectorized a function. If you look on the issue (https://github.com/stan-dev/math/issues/202) and on Bob's first comment, I've been checking off which functions I have vectorized. The issues I've created are the ones that I haven't been able to vectorize because it's missing a fwd or rev mode.

Best,

Rayleigh

--
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.

[Daniel Lee]

I haven't seen a single pull request go through. Are you batching them all? If so, mind submitting a pull request for one of the functions so we can see what it's going to look like?

Daniel

[Bob Carpenter]

Or just submit a request for all the ones that work so far.
Each one should be pretty simple.

- Bob

[Bob Carpenter]

As I mentioned before, each function will need a Stan program
to test its instantiation. You can update the existing test models
in stan/src/test/test-models/good/function-signatures/

- Bob

[Rayleigh L]

If you look on the branch 'issue-202/vectorize-all', I've vectorized all the functions that don't have other issues. I still have to document them, but there are 118 changed files. So, in hindsight, I probably should have clarified how to stage these changes first. However, each commit on the branch except for the one including the broken exp.hpp should contain a subset of working vectorized functions.

[Bob Carpenter]

I think doing them all at once for this is OK.

- Bob