When sampling bounded variables using NUTS, does pyMC first sample in terms of a continuous variable transform, allowing for well behaved calculations of the gradient, and then map these samples to the bounded domain? If so when using my own custom likelihood function with bound random variable pri…

Bounded Variables and Custom Likelihood Functions

aarcher June 6, 2022, 3:12pm 3

Great! Thank you so much for your detailed response!

Topic		Replies	Views
Bounded variable transformation when not using HMC? version agnostic	8	628	February 14, 2023
Bounded variable logp Questions	6	902	August 15, 2018
X_lowerbound__ or X_interval__ in the Trace Questions	1	701	November 21, 2017
NUTS - FromFunctionOp has no attribute 'grad' Questions theano	2	1005	August 3, 2018
Prior boundaries that are not constant but random variables Questions	6	901	December 14, 2017