Hi,

I’m having problem trying to access the logp function of a normal mixture.

This first attempt, let me think that I understand how to access the logp function of a basic random variable :

```
with pm.Model() as model:
test = pm.Normal('test',mu=0,sd=1)
lp = model.logp({'test':12})
```

which gives lp =~ -73 (which sounds ok).

But I have a problem when trying to do the same stuff with a little more complex random variable, a normal mixture :

```
N = 5
with pm.Model() as model:
w = pm.Dirichlet('w', a=np.ones(N))
mu = pm.Normal('mu',mu=0,sd=1,shape=N)
sd = pm.HalfNormal('sd',sd=1,shape=N)
mix = pm.NormalMixture('mix',w=w,mu=mu, sd=sd) # building the mixture
lp = model.logp({'mix':12}) # logp call
```

This gives me the following error : “*TypeError: Missing required input: w_stickbreaking__*” linked to the *logp* call…

I really do not understand why this doesn’t work: I see no functionnal difference between the two codes samples.

And the error message seem to talk about a missing feature w, which is explicitely given !! (Note : the code works without the *logp* call)

Any help will be appreciated.