How to increase posterior sample count when fitting GP model with find_MAP()?

Yes, alter the InferenceData/trace/dict returned by find_MAP to make it look like it had N draws (all identical). Then call posterior predictive on that altered trace

2 Likes