is it possible to code multivariate outcomes in map2stan?

[rmcelreath]

is it possible to code multivariate outcomes in map2stan?

[rmcelreath]

This is not supported yet, no.

I had started to add this use case, but then stopped when I decided I should refactor the way map2stan handles distributions, so that users can more easily add their own.

This is a good reminder to do something about that.

[andersgs]

Thank you for the quick reply. I would be more than happy to help in refactoring map2stan.

[mbeddo]

I am helping a college perform a student retention study. I wanted to use some ideas from chapter 13 because I believe some of the variables are correlated. But I can't get my model to work with map2stan, and I can't make the examples in the chapter work either.

I get "Rho : no dimension found for this matrix. Use explicit start value to define its dimension. For example: Rho = diag(2)". I've tried dmvnorm, dmvnorm2, and dmvnormNC variates.

So if I include "start = list(Rho=diag(2))" in the arguments to map2stan, I then get "Error in paste(..., collapse = "", sep = "") : argument "n" is missing, with no default"

I use R version 3.2.4 Revised (2016-03-16 r70336), and rstan (Version 2.11.1, packaged: 2016-07-28 18:19:31 UTC, GitRev: 85f7a56811da)

[rmcelreath]

mbeddo, your issue doesn't belong in this thread. Please start another issue and include your code.