Hi there everyone, I am a little confused about how “logp” works for model object with pymc3. I have a simple model for dataset y_1, \dots, y_n given by y \sim GP(0, k_\theta(x,x')) . I am trying to approximate the integral p(y_1, \dots, y_n) = \int_\theta p(y_1, \dots, y_n|\theta)p(\theta)d\theta …

GP and MC integral approximation - which logp to use?

kejzlarv February 11, 2019, 10:56pm 2

I see, based on https://docs.pymc.io/developer_guide.html#logp-and-dlogp, y_gp_1.logp is the one I am looking for. Am I right?

Topic		Replies	Views
How to calculate log posterior of a GP over a trace in pymc3 Questions	14	2287	February 2, 2018
Pymc3.model likelihood evaluation for given prior sample Questions	2	550	June 22, 2018
Evaluate logposterior at sample points Questions	9	4411	August 16, 2017
Likelihood for the product of two Gaussian random variables Questions	2	1189	February 16, 2018
Compute the KL divergence between two distributions Questions	12	2680	August 13, 2018