return (conditional) covariance matrix in `predict` function
Closed this issue · 4 comments
currently only the diagonal elements are returned, and to support simulate method, the full matrix is needed.
while implementing this, a question comes to my mind: is the simulate method, giving simulations on
I think we should give the latter one, what do you think @danielinteractive ?
idea: conduct the simulation based on the covariance in C++ directly, instead of returning the list of covariance matrix for each subject.
Hm - it could be beneficial to do the simulation in R to preserve readability for R developers?
while implementing this, a question comes to my mind: is the simulate method, giving simulations on θ, or (θ and β)?
I think we should give the latter one, what do you think @danielinteractive ?
good question - let's work on this in #308