Errors using sample_posterior_predictive

[paulstillman]

I'm not positive if this is the right place for this (or if I should open an issue), but I'm encountering errors when trying to get the sample_posterior_predictive to work. Below is a reproducible example (fails on both 3.9.17 and 3.10.12, output is from the latter):

import hssm
hssm.set_floatX("float32")
v_true, a_true, z_true, t_true = [0.5, 1.5, 0.5, 0.5]
dataset = hssm.simulate_data(
    model="ddm",
    theta=[v_true, a_true, z_true, t_true],
    size=1000,
)
simple_ddm_model = hssm.HSSM(data=dataset)
simple_ddm_model.sample()
simple_ddm_model.sample_posterior_predictive()

The last line outputs the following two exceptions:

Traceback (most recent call last):
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/pytensor/compile/function/types.py", line 970, in __call__
    self.vm()
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/pytensor/graph/op.py", line 552, in rval
    r = p(n, [x[0] for x in i], o)
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/pytensor/tensor/random/op.py", line 333, in perform
    smpl_val = self.rng_fn(rng, *(args + [size]))
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/hssm/distribution_utils/dist.py", line 255, in rng_fn
    elif size % new_data_size != 0:
TypeError: unsupported operand type(s) for %: 'NoneType' and 'int'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/hssm/hssm.py", line 608, in sample_posterior_predictive
    return self.model.predict(idata, kind, data, inplace, include_group_specific)
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/bambi/models.py", line 857, in predict
    pps = self.family.posterior_predictive(**required_kwargs, **optional_kwargs)
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/bambi/families/family.py", line 184, in posterior_predictive
    output_array = pm.draw(response_dist.dist(**kwargs))
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/pymc/sampling/forward.py", line 322, in draw
    return draw_fn()
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/pytensor/compile/function/types.py", line 983, in __call__
    raise_with_op(
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/pytensor/link/utils.py", line 535, in raise_with_op
    raise exc_value.with_traceback(exc_trace)
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/pytensor/compile/function/types.py", line 970, in __call__
    self.vm()
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/pytensor/graph/op.py", line 552, in rval
    r = p(n, [x[0] for x in i], o)
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/pytensor/tensor/random/op.py", line 333, in perform
    smpl_val = self.rng_fn(rng, *(args + [size]))
  File "/Users/paulstillman/opt/anaconda3/envs/py10/lib/python3.10/site-packages/hssm/distribution_utils/dist.py", line 255, in rng_fn
    elif size % new_data_size != 0:
TypeError: unsupported operand type(s) for %: 'NoneType' and 'int'
Apply node that caused the error: SSM_RV_rv{1, (0, 0, 0, 0, 0), floatX, True}(RandomGeneratorSharedVariable(<Generator(PCG64) at 0x16D7F5E00>), [], 10, [[[0.49938 ... 5140317]]], [[[1.41082 ... 4278388]]], [[[0.48220 ... 949423 ]]], [[[0.54447 ... 3136015]]], [[[0.05]]])
Toposort index: 0
Inputs types: [RandomGeneratorType, TensorType(int64, shape=(0,)), TensorType(int64, shape=()), TensorType(float32, shape=(4, 1000, 1000)), TensorType(float32, shape=(4, 1000, 1)), TensorType(float32, shape=(4, 1000, 1)), TensorType(float32, shape=(4, 1000, 1)), TensorType(float32, shape=(1, 1, 1))]
Inputs shapes: ['No shapes', (0,), (), (4, 1000, 1000), (4, 1000, 1), (4, 1000, 1), (4, 1000, 1), (1, 1, 1)]
Inputs strides: ['No strides', (0,), (), (4000, 4, 16000), (4000, 4, 4), (4000, 4, 4), (4000, 4, 4), (4, 4, 4)]
Inputs values: [Generator(PCG64) at 0x16D7F5E00, array([], dtype=int64), array(10), 'not shown', 'not shown', 'not shown', 'not shown', array([[[0.05]]], dtype=float32)]
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.

I'm new to both hssm and hddm (loving hssm so far!) so it's definitely possible that my inexperience is making me miss something obvious

[digicosmos86]

Hi Paul,

Thanks for letting us know about this! It does seem to be a bug on our end. We will look into it and hopefully fix it in our next update

[paulstillman]

Thank you!