I want to model a matrix as a multivariate normal with univariate normal priors on the column means and an LKJ prior on the covariance matrix. Something like this works fine, as expected: df = (df - df.mean()) / df.std() # necessary! m = len(df.columns) mu = pm.Normal("mu", 0, sd=1, shape=m) sd_di…

Using multivariate normal with widely varying means

junpenglao October 1, 2019, 10:49am 2

did you try to set the test val of mu? It might help but I am not sure how much.

Topic		Replies	Views
Overcoming divergence in MvNormal covariance model? Questions	1	481	January 28, 2019
About set of Multivariate Normal Distributions Questions	2	1187	July 3, 2018
Multivariate normal distribution Questions	4	2617	March 29, 2019
Hyperpriors for a MvNormal prior Questions	3	637	February 18, 2021
Using a Multivariate Normal as a prior Questions	1	1171	March 4, 2019