Per the discussion at the meeting today I wrote down the required function implementations for adding a Gaussian process in the Stan Math Library, https://github.com/stan-dev/stan/wiki/Adding-a-Gaussian-Process-Covariance-Function.
I tried to keep the same naming prefix convention that the
cov_exp_quad function uses, but that would lead to inconsistencies with suffix notation the log density functions use so I went with the latter. Hopefully this won’t be too much of a conflict, as we can always keep
cov_exp_quad around and deprecate it when we introduce
@Lu.Zhang, eventually we’d want your NNGP to be implemented with this pattern so that it will be easy for our users to incorporate into their models. Does it make sense or is there anything on which I should elaborate? Thanks!