Aki Vehtari

I emailed Lu Cheng

Thanks for the report. It would be great if you have time to fix this. Otherwise, I can fix this in July.

Thanks for the suggestion.

Different R packages use acronym `gpd`. Python SciPy uses `genpareto` I'm used to `gpd`, but if you feel `gen_pareto` is more clear, then I'm fine with that. > Presumably we'd...

This is part of the path towards INLA speed for Gaussian latent variable models in Stan. Anticipated first use case is compound functions for GLMs with Laplace integration over weights....

> shouldn't we have a different sigma for the intercept and all other regressors? Sure. I didn't say that sigma would be a scalar. We'll make a more detail proposal...

I edited the issue title and suggest adding also `weibull_log_glm_*` which is very common in survival analysis. In survival analysis with censored observations, it would be very useful to have...

If x is augmented with column of 1's, the first sigma can be included in Sigma. Now Sigma is scalar. The request is to allow it to be a diagonal...

> k(x1, x2) = x1^T diag(sigma^2) x2? I'm fine with this (just don't explicitly create the diagonal matrix)

All _cov things are currently just creating matrices, so we can go with in this case, too. Right now as mentioned in the issue, I'm hoping just the modification that...