ValueError : Zeros Dimensional Array in pm.traceplot

[garnier94]

I'm trying to reproduce the code used in the AR(2) example. The sampling is normal but I receive a ValueError when i try to use pm.traceplot. I think that this issue is related to the way arviz handles MultiTrace.

Description of your problem

Please provide a minimal, self-contained, and reproducible example.

import pymc3 as pm
import numpy as np

np.random.seed(seed=1848)
T = 100
y = np.zeros((T,))
for i in range(1,T):
    y[i] = 0.95 * y[i-1] - 0.05 * y[i-2]+ np.random.normal()

tau = 1.0
with pm.Model() as ar2:
    beta = pm.Normal('beta', mu=0, sigma=tau, shape=2)    
    data = pm.AR('y', beta, sigma=1, observed=y)
    trace = pm.sample(1000, cores=4)

pm.traceplot(trace)

Please provide the full traceback.

<ipython-input-4-dc52240121b7> in <module>()
----> 1 pm.traceplot(trace)

*/python3.6/site-packages/pymc3/plots/__init__.py in wrapped(*args, **kwargs)
     40                 warnings.warn('Keyword argument `{old}` renamed to `{new}`, and will be removed in pymc3 3.8'.format(old=old, new=new))
     41                 kwargs[new] = kwargs.pop(old)
---> 42             return func(*args, **kwargs)
     43     return wrapped
     44 

*/python3.6/site-packages/arviz/plots/traceplot.py in plot_trace(data, var_names, coords, divergences, figsize, textsize, lines, combined, plot_kwargs, fill_kwargs, rug_kwargs, hist_kwargs, trace_kwargs)
    102             divergences = False
    103 
--> 104     data = convert_to_dataset(data, group="posterior")
    105     var_names = _var_names(var_names, data)
    106 

*/python3.6/site-packages/arviz/data/converters.py in convert_to_dataset(obj, group, coords, dims)
    156     xarray.Dataset
    157     
--> 158     inference_data = convert_to_inference_data(obj, group=group, coords=coords, dims=dims)
    159     dataset = getattr(inference_data, group, None)
    160     if dataset is None:

*/python3.6/site-packages/arviz/data/converters.py in convert_to_inference_data(obj, group, coords, dims, **kwargs)
     80             return from_pystan(**kwargs)
     81     elif obj.__class__.__name__ == "MultiTrace":  # ugly, but doesn't make PyMC3 a requirement
---> 82         return from_pymc3(trace=kwargs.pop(group), **kwargs)
     83     elif obj.__class__.__name__ == "EnsembleSampler":  # ugly, but doesn't make emcee a requirement
     84         return from_emcee(sampler=kwargs.pop(group), **kwargs)

*/python3.6/site-packages/arviz/data/io_pymc3.py in from_pymc3(trace, prior, posterior_predictive, coords, dims)
    150         posterior_predictive=posterior_predictive,
    151         coords=coords,
--> 152         dims=dims,
    153     ).to_inference_data()

*/python3.6/site-packages/arviz/data/io_pymc3.py in to_inference_data(self)
    135             **{
    136                 "posterior": self.posterior_to_xarray(),
--> 137                 "sample_stats": self.sample_stats_to_xarray(),
    138                 "posterior_predictive": self.posterior_predictive_to_xarray(),
    139                 "prior": self.prior_to_xarray(),

*/python3.6/site-packages/arviz/data/base.py in wrapped(cls, *args, **kwargs)
     23                 if getattr(cls, prop) is None:
     24                     return None
---> 25             return func(cls, *args, **kwargs)
     26 
     27         return wrapped

*/python3.6/site-packages/arviz/data/io_pymc3.py in sample_stats_to_xarray(self)
     76             name = rename_key.get(stat, stat)
     77             data[name] = np.array(self.trace.get_sampler_stats(stat, combine=False))
---> 78         log_likelihood, dims = self._extract_log_likelihood()
     79         if log_likelihood is not None:
     80             data["log_likelihood"] = log_likelihood

*/python3.6/site-packages/arviz/data/base.py in wrapped(cls, *args, **kwargs)
     23                 if getattr(cls, prop) is None:
     24                     return None
---> 25             return func(cls, *args, **kwargs)
     26 
     27         return wrapped

*/python3.6/site-packages/arviz/data/io_pymc3.py in _extract_log_likelihood(self)
     54         for chain in self.trace.chains:
     55             log_like = (log_likelihood_vals_point(point) for point in self.trace.points([chain]))
---> 56             chain_likelihoods.append(np.stack(log_like))
     57         return np.stack(chain_likelihoods), coord_name
     58 

*/python3.6/site-packages/numpy/core/shape_base.py in stack(arrays, axis, out)
    408     """
    409     _warn_for_nonsequence(arrays)
--> 410     arrays = [asanyarray(arr) for arr in arrays]
    411     if not arrays:
    412         raise ValueError('need at least one array to stack')

*/python3.6/site-packages/numpy/core/shape_base.py in <listcomp>(.0)
    408     """
    409     _warn_for_nonsequence(arrays)
--> 410     arrays = [asanyarray(arr) for arr in arrays]
    411     if not arrays:
    412         raise ValueError('need at least one array to stack')

*/python3.6/site-packages/arviz/data/io_pymc3.py in <genexpr>(.0)
     53         chain_likelihoods = []
     54         for chain in self.trace.chains:
---> 55             log_like = (log_likelihood_vals_point(point) for point in self.trace.points([chain]))
     56             chain_likelihoods.append(np.stack(log_like))
     57         return np.stack(chain_likelihoods), coord_name

*/python3.6/site-packages/arviz/data/io_pymc3.py in log_likelihood_vals_point(point)
     49                     log_like_val = log_like_val[~var.observations.mask]
     50                 log_like_vals.append(log_like_val)
---> 51             return np.concatenate(log_like_vals)
     52 
     53         chain_likelihoods = []

ValueError: zero-dimensional arrays cannot be concatenated

Please provide any additional information below.

This problem doesn't depend on the numpy version (even if i get a FutureWarning on the behavior of concatenation). I'm using numpy 1.16.4, but i still have the issue with older version of numpy.
This problem also happens in the AR(1) case, when i use shape = 1 .

Versions and main components

• PyMC3 Version: 3.7
• Theano Version: 1.0.4
• Python Version: 3.6.1
• Arviz Version : 0.4.0
• Operating system: Debian 3.16.43
• How did you install PyMC3: pip

[aloctavodia]

This have been addressed on the ArviZ side arviz-devs/arviz#693

[lucianopaz]

Closing thanks to the ArviZ fix. Installing the bleeding edge version of ArviZ from their github repo will fix the problem.