Error in prior predictive with Gamma distribution as Likelihood

1

This code runs smoothly,

with pm.Model() as model_test:
    slope_ = pm.Normal("slope", mu=0, sigma=1)
    intercept_ = pm.Normal("intercept", mu = 3, sigma=1)
    sigma_ = pm.HalfNormal("sigma", sigma = 1)
    mu = slope_*x + intercept_
    y_est = pm.Normal("y_est", mu=mu, sigma=sigma_, observed=y) 
    spc = pm.sample_prior_predictive(samples=100)

But this one raises an strange PyTensor error,

with pm.Model() as model_test:
    slope_ = pm.Normal("slope", mu=0, sigma=1)
    intercept_ = pm.Normal("intercept", mu = 3, sigma=1)
    sigma_ = pm.HalfNormal("sigma", sigma = 1)
    mu = slope_*x + intercept_
    y_est = pm.Gamma("y_est", mu=mu, sigma=sigma_, observed=y)  
    spc = pm.sample_prior_predictive(samples=100)

The error:

in Function.call(self, *args, **kwargs) 968 try: 969 outputs = ( → 970 self.vm() 971 if output_subset is None 972 else self.vm(output_subset=output_subset) 973 ) 974 except Exception: in Op.make_py_thunk…rval(p, i, o, n, params) 539 @is_thunk_type 540 def rval( 541 p=p, i=node_input_storage, o=node_output_storage, n=node, params=None 542 ): → 543 r = p(n, [x[0] for x in i], o) 544 for o in node.outputs: File [~/miniconda3/envs/PyMC5_2/lib/python3.10/site-packages/pytensor/tensor/random/op.py:378]lib/python3.10/site-packages/pytensor/tensor/random/op.py:378), in RandomVariable.perform(self, node, inputs, outputs) 376 rng_var_out[0] = rng → 378 smpl_val = self.rng_fn(rng, *(args + [size])) 380 if ( 381 not isinstance(smpl_val, np.ndarray) 382 or str(smpl_val.dtype) != out_var.type.dtype
…

Inputs values: [Generator(PCG64) at 0x7F2C51526DC0, array([87]), array(11), ‘not shown’, ‘not shown’] Outputs clients: [[‘output’], [‘output’]]

HINT: Re-running with most PyTensor optimizations disabled could provide a back-trace showing when this node was created. This can be done by setting the PyTensor flag ‘optimizer=fast_compile’. If that does not work, PyTensor optimizations can be disabled with ‘optimizer=None’.
HINT: Use the PyTensor flag exception_verbosity=high for a debug print-out and storage map footprint of this Apply node.

2

I assume this is because slope_*x + intercept_ is sometimes negative. Have you checked that with your x values?

3

Ahhh… of course you are right. It shouldn’t be negative but it can be. I will check and come back.

1 Like