Hi all, I am trying to figure out what does the NUTS sampler do when it is not given an observed data. Intuitively I’d expect it will work like just a regular numpy random number generator, but this seems not to be the case. Here is a quick demo: joint_model = pm.Model() var = np.full(400,1e-6) va…

When you call pm.sample it uses MCMC to sample parameter values. The specific MCMC algorithm you get by default requires some tuning to work properly so limiting this stage like you did hurts convergence. If you use pm.sample_prior_predictive instead it will work like a simple numpy random function…

Sampling without the observed keyword

Questions

AlanZ November 2, 2022, 5:15pm 3

Thank you so much! This is exactly what I need!

Topic		Replies	Views
Sample method of pymc3 Questions	4	685	July 15, 2018
Just using NUTS Questions	7	1168	May 5, 2018
Very new to all of this. Why doesn't this three line toy example work? Questions	3	527	June 25, 2019
Differenc between PyMC 3.1 and PyMC 3.0 on NUTS Questions	1	1297	August 28, 2017
Is there any material gently explain how PYMC3 class/function works Questions	2	648	June 29, 2019

Sampling without the observed keyword

Related topics