I am putting together a network latent space model, but can’t get it to sample in pymc3. I’m afraid I’m missing something obvious to do either with using theano transformations or shape parameters, but can’t figure out which. Here is the simplest version of the model that replicates the problem. N …

Yeah this is not very clear from the doc but passing step=... to pm.sample() inhibits the init=..., and as the default value of z is all 0 that’s the reason for Bad initial energy... error. The above code I posted with non-zero testval should works.

No sampling when transforming random variables with theano using dot product

Questions

lwahedi February 8, 2019, 4:21pm 10

Thanks, that clears it up, and will hopefully help with sampling issues in the future.

Topic		Replies	Views
ERROR (theano.gof.opt) & slowness in execution in sampling Questions	3	1168	March 27, 2019
Sampling issue: The following error happened while compiling the node Questions theano	4	3294	January 21, 2019
Why isn't NUTS sampling with tt.dot or pm.math.dot? Questions	5	1294	October 7, 2018
Trouble with Initilization and Posterior Sampling Questions theano	13	2225	June 15, 2018
pm.Potential without Theano? Questions	14	2301	March 20, 2018

No sampling when transforming random variables with theano using dot product

Related topics