Hi! I have the following problem and couldn’t find any way to debug it further. When I run the code: with pm.Model() as model: B1 = pm.Lognormal('B1', -2, 2) B1_print = tt.printing.Print('B1')(B1) jm.B1 = B1_print image_model = jm.image() image_obs = pm.Normal('image_obs', mu=im…

Sampler (NUTS) proposes the same value

aplavin April 26, 2018, 9:58pm 9

It turned out that the issue has something to do with theano broadcasting, and I managed to fix it (even through not completely understanding).

Topic		Replies	Views
Sampling time is very long (minutes per sample) Questions	13	4561	November 2, 2017
Runtime Exception with NUTS v3	16	1137	March 16, 2022
Switch point Metropolis tuning Questions	25	5000	May 19, 2018
Trouble with Initilization and Posterior Sampling Questions theano	13	2225	June 15, 2018
Averaging over model output that sometimes returns null values? Questions	12	1458	November 16, 2017