Call MvNormal.random() method inside model

Questions
2

[but why].gif

Joke aside, really why would you want to do that? Passing random generation from one of the node to downstream as observed sounds like a bad idea.

Since .random() wont be directly called in a pm.Model, if you really want you can generate it from theano random generator (already wrapped in pymc3):

import pymc3 as pm
from pymc3 import tt_rng

n = 10000 # sample size
m = 5 # number of covariates

with pm.Model() as model:
    packed_L = pm.LKJCholeskyCov('packed_L', n=m, eta=2., sd_dist=pm.HalfCauchy.dist(2.5))
    L = pm.expand_packed_triangular(m, packed_L)
    Sigma = pm.Deterministic('Sigma', L.dot(L.T))
    mu = pm.Normal('mu', mu=0., sd=1., shape=m)
    Y = pm.MvNormal('Y', mu=mu, chol=L, shape=(n,m))
    sample = mu + tt_rng().normal(size=(n,m)).dot(L)
1 Like