bgoodri

Yes, although we might as well implement one of the generalizations of the generalized gamma distribution. On Fri, Jul 10, 2015 at 1:20 PM, Bob Carpenter [email protected] wrote: > @steven-murray...

Perhaps. Do utilize the `expose_stan_functions` function in rstan 2.7.x (now on CRAN) to verify that your implementation agrees with GSL's via the RcppGSL or gsl packages.

Well, Andrew was trying to use this in a ML context, so if we can only get it working for that first, it would be okay. The generated quantities block...

You do need the score on an observation-by-observation basis to estimate the Hessian this way. Maybe there is some other way to expose the Hessian to the generated quantities block....

On Wed, Nov 30, 2016 at 10:23 AM, Daniel Lee wrote: > @bgoodri , are you envisioning something like > this only happening in generated quantities? No; the likelihood can...

On Wed, Nov 30, 2016 at 11:11 AM, Daniel Lee wrote: > Hmm... that makes it much more of a pain, but I don't think it's > impossible. We'd need...

This is going to be fine. We need to test whether it continues to work on older Windows releases of R and if not, bump the minimum R version up...

Yes, if we enabled LTO for many, many other packages via rstantools, then we would generate the package's .def file when the `configure.win` script is called (which in turn calls...

OK, this is weird. It seems that we now have the `m` not found problem from `.onLoad` even if I remove the LTO stuff from src/Makevars.win; see [here](https://win-builder.r-project.org/zF29gTB9s3DT/). But it...

rstan 2.32 is going to CRAN today, which will fix the test failures with `clang++16`