How can we build a mixture of mixtures?

Questions
10

Oh OK, I think I might have lead you to a wrong path :sweat_smile:
In your case, you can use pm.Mixture out of the box:

nbr = 5
with pm.Model() as model:
    w = pm.Dirichlet('w', a=np.array([0.0000001]*nbr)) # weight vector for the mixture
    mu = pm.Exponential('mu', lam=1, shape=nbr)
    sd = pm.Exponential('sd', lam=1, shape=nbr)
    components = pm.Lognormal.dist(mu=mu, sd=sd, shape=nbr) 
    obs = pm.Mixture('y', w, components, observed=data)