I was trying out the examples in the book : Probabilistic-Programming-and-Bayesian-Methods-for-Hackers.
Towards the end of chapter 2 I got stuck at the block of code below.
It purpose is to determine if a simulated data from a posterior distribution matches the original data, the code block is the first step in the process.
I am using pymc not pymc3.
CODE BLOCK:
N = 10000
with pm.Model() as model:
beta = pm.Normal("beta", mu=0, tau=0.001, testval=0)
alpha = pm.Normal("alpha", mu=0, tau=0.001, testval=0)
p = pm.Deterministic("p", 1.0/(1. + tt.exp(beta*temperature + alpha)))
observed = pm.Bernoulli("bernoulli_obs", p, observed=D)
# pytensor.config.compute_test_value = "warn"
simulated = pm.Bernoulli("bernoulli_sim", p,shape= p.shape )
step = pm.Metropolis(vars=[p])
trace = pm.sample(N, step=step)
ERROR:
ValueError Traceback (most recent call last)
Cell In[32], line 10
7 # pytensor.config.compute_test_value = "warn"
9 simulated = pm.Bernoulli("bernoulli_sim", p,shape= p.shape )
---> 10 step = pm.Metropolis(vars=[p])
11 trace = pm.sample(N, step=step)
File ~\.conda\envs\bayes_course\Lib\site-packages\pymc\step_methods\metropolis.py:168, in Metropolis.__init__(self, vars, S, proposal_dist, scaling, tune, tune_interval, model, mode, **kwargs)
166 vars = model.value_vars
167 else:
--> 168 vars = get_value_vars_from_user_vars(vars, model)
170 initial_values_shape = [initial_values[v.name].shape for v in vars]
171 if S is None:
File ~\.conda\envs\bayes_course\Lib\site-packages\pymc\util.py:480, in get_value_vars_from_user_vars(vars, model)
477 notin = list(map(get_var_name, notin))
478 # We mention random variables, even though the input may be a wrong value variable
479 # because most users don't know about that duality
--> 480 raise ValueError(
481 "The following variables are not random variables in the model: " + str(notin)
482 )
484 return value_vars
ValueError: The following variables are not random variables in the model: ['p']
I don't believe you need to provide p to Metropolis. It should be able to use the deterministic var through Bernoulli where it was provided.
From the documentation here:
Deterministic quantities are just computeed once at the end of the step, with the final values of the other random variables
This sounds to me like you can remove vars=[p] and use
step = pm.Metropolis()
directly.
I tried it locally with some fake data and it appears that p is in the final trace.
print(trace.posterior)
Data variables:
temperature (chain, draw) float64 -62.98 -21.79 -13.69 ... 18.63 18.63
beta (chain, draw) float64 -26.23 -26.23 -39.61 ... 25.9 38.93
alpha (chain, draw) float64 -53.5 -53.5 -5.149 ... -0.5374 -0.5374
bernoulli_sim (chain, draw) int64 0 0 0 0 0 0 0 0 0 0 ... 0 0 0 0 0 0 0 0 0
p (chain, draw) float64 0.0 1.022e-225 ... 4.518e-210 0.0