jtimonen/lgpr

Easier/faster prior predictive checks

jtimonen opened this issue · 1 comments

The current way to do a prior predictive check is to first run lgp() using the options sample_f=TRUE and prior_only=TRUE after which one can use ppc(). This is not any faster than posterior sampling, and it would probably be much faster to just draw from
the hyperparameter priors separately first, and then for each draw use lgp() with sample_f=TRUE, prior_only=TRUE and algorithm="Fixed_param".

Version 1.1 now has prior_pred() and sample_param_prior().