lkj cov distribution tests

[Peter Li]

Are there any tests for LKJ cov? In src/test/unit/prob/distributions/multivariate/continuous/lkj_cov_test.cpp, none of the tests call lkj_cov, they call lkj_corr. I'm trying to add tests for the higher order autodiff to make sure fvars can be passed in but I don't see anything to base the tests off of.

[Ben Goodrich]

On Monday, August 11, 2014 11:14:39 AM UTC-4, Peter Li wrote:

Are there any tests for LKJ cov? In src/test/unit/prob/distributions/multivariate/continuous/lkj_cov_test.cpp, none of the tests call lkj_cov, they call lkj_corr. I'm trying to add tests for the higher order autodiff to make sure fvars can be passed in but I don't see anything to base the tests off of.

I think we would be better off getting rid of lkj_cov because it is a combination of a lognormal prior on the standard deviations and a LKJ prior on the correlation matrix that users could just as easily do themselves with any other reasonable prior for the standard deviations. And it takes a whole correlation matrix when it would be better if the model were parameterized in terms of L. But if lkj_cov survives, then we should make sure it is tested.

Ben

[Bob Carpenter]

I'm happy scheduling it for retirement. Should we deprecate
it first or just trash it? Removing it breaks backward compatibility,
though I don't know if anyone's even using this density.

- Bob

> --
> 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.

[Ben Goodrich]

I would say break it for Stan 3.0 but maybe deprecate it if we have another 2.x release. I doubt very many people are using it.

> To unsubscribe from this group and stop receiving emails from it, send an email to stan-dev+unsubscribe@googlegroups.com.

[Bob Carpenter]

Will do:

https://github.com/stan-dev/stan/issues/865

I assigned the issue to myself.

- Bob

> > 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.
>
>

> --
> 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.