Can NUTS help me to find a set of parameters with largest likelihood if my model is identifiable only up to rearrangement? One of the parameter is a matrix \mathcal{A}. Let’s say matrix A_{1} is a possible value of \mathcal{A}. I can form a new matrix A_{2} by the following two steps: swap i^{\t…

[image] Hamster_on_wheels: Would the identifiability problem mess up the tuning of the mass matrix and NUTS? I would say no as the mass matrix estimation is over the whole posterior, so label switching wont effect that [image] Hamster_on_wheels: What is “within trace mode exploring”? I…

Identifiable up to rearrangement

Questions

Hamster_on_wheels May 29, 2019, 3:29pm 6

Would the identifiability problem mess up the tuning of the mass matrix and NUTS?
What is “within trace mode exploring”?

Topic		Replies	Views
Divergences and label-switching in a Dirichlet process mixture model Questions	2	1334	July 18, 2021
Label switching in multivariate mixtures Questions	0	765	September 28, 2020
Constraining order in a Mixture model Questions	6	1440	August 9, 2021
Constrain one variable to be greater than another Questions	14	3519	January 30, 2018
Why does `transform=pm.distributions.transforms.ordered` lead to worse convergence? Questions	3	1579	August 27, 2021

Identifiable up to rearrangement

Related topics