Theano import issue

[MaxHalford]

Description of your problem

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

import pymc3 as pm

with pm.Model() as model:
    
    x = pm.Uniform('x', lower=0, upper=16)

Please provide the full traceback.

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-4-5fea95be487c> in <module>()
      3 with pm.Model() as model:
      4 
----> 5     x = pm.Uniform('x', lower=0, upper=16)

~/anaconda3/lib/python3.6/site-packages/pymc3/distributions/distribution.py in __new__(cls, name, *args, **kwargs)
     39                 raise TypeError("observed needs to be data but got: {}".format(type(data)))
     40             total_size = kwargs.pop('total_size', None)
---> 41             dist = cls.dist(*args, **kwargs)
     42             return model.Var(name, dist, data, total_size)
     43         else:

~/anaconda3/lib/python3.6/site-packages/pymc3/distributions/distribution.py in dist(cls, *args, **kwargs)
     50     def dist(cls, *args, **kwargs):
     51         dist = object.__new__(cls)
---> 52         dist.__init__(*args, **kwargs)
     53         return dist
     54 

~/anaconda3/lib/python3.6/site-packages/pymc3/distributions/continuous.py in __init__(self, lower, upper, *args, **kwargs)
    181         self.lower = lower = tt.as_tensor_variable(floatX(lower))
    182         self.upper = upper = tt.as_tensor_variable(floatX(upper))
--> 183         self.mean = (upper + lower) / 2.
    184         self.median = self.mean
    185 

~/anaconda3/lib/python3.6/site-packages/theano/tensor/var.py in __add__(self, other)
    126     def __add__(self, other):
    127         try:
--> 128             return theano.tensor.basic.add(self, other)
    129         # We should catch the minimum number of exception here.
    130         # Otherwise this will convert error when Theano flags

~/anaconda3/lib/python3.6/site-packages/theano/gof/op.py in __call__(self, *inputs, **kwargs)
    668                 # compute output value once with test inputs to validate graph
    669                 thunk = node.op.make_thunk(node, storage_map, compute_map,
--> 670                                            no_recycling=[])
    671                 thunk.inputs = [storage_map[v] for v in node.inputs]
    672                 thunk.outputs = [storage_map[v] for v in node.outputs]

~/anaconda3/lib/python3.6/site-packages/theano/gof/op.py in make_thunk(self, node, storage_map, compute_map, no_recycling, impl)
    953             try:
    954                 return self.make_c_thunk(node, storage_map, compute_map,
--> 955                                          no_recycling)
    956             except (NotImplementedError, utils.MethodNotDefined):
    957                 # We requested the c code, so don't catch the error.

~/anaconda3/lib/python3.6/site-packages/theano/gof/op.py in make_c_thunk(self, node, storage_map, compute_map, no_recycling)
    856         _logger.debug('Trying CLinker.make_thunk')
    857         outputs = cl.make_thunk(input_storage=node_input_storage,
--> 858                                 output_storage=node_output_storage)
    859         thunk, node_input_filters, node_output_filters = outputs
    860 

~/anaconda3/lib/python3.6/site-packages/theano/gof/cc.py in make_thunk(self, input_storage, output_storage, storage_map, keep_lock)
   1215         cthunk, module, in_storage, out_storage, error_storage = self.__compile__(
   1216             input_storage, output_storage, storage_map,
-> 1217             keep_lock=keep_lock)
   1218 
   1219         res = _CThunk(cthunk, init_tasks, tasks, error_storage, module)

~/anaconda3/lib/python3.6/site-packages/theano/gof/cc.py in __compile__(self, input_storage, output_storage, storage_map, keep_lock)
   1155                                             output_storage,
   1156                                             storage_map,
-> 1157                                             keep_lock=keep_lock)
   1158         return (thunk,
   1159                 module,

~/anaconda3/lib/python3.6/site-packages/theano/gof/cc.py in cthunk_factory(self, error_storage, in_storage, out_storage, storage_map, keep_lock)
   1607         """
   1608         try:
-> 1609             key = self.cmodule_key()
   1610         except KeyError:
   1611             key = None

~/anaconda3/lib/python3.6/site-packages/theano/gof/cc.py in cmodule_key(self)
   1298                                  libraries=self.libraries(),
   1299                                  header_dirs=self.header_dirs(),
-> 1300                                  c_compiler=self.c_compiler(),
   1301                                  )
   1302 

~/anaconda3/lib/python3.6/site-packages/theano/gof/cc.py in cmodule_key_(self, fgraph, no_recycling, compile_args, libraries, header_dirs, insert_config_hash, c_compiler)
   1377         # DynamicModule always add the include <numpy/arrayobject.h>
   1378         sig.append('NPY_ABI_VERSION=0x%X' %
-> 1379                    np.core.multiarray._get_ndarray_c_version())
   1380         if c_compiler:
   1381             sig.append('c_compiler_str=' + c_compiler.version_str())

AttributeError: module 'numpy.core.multiarray' has no attribute '_get_ndarray_c_version'

Please provide any additional information below.

I tried to create a fresh virtual environment and reinstall different versions of Theano but nothing worked. Also I'm working in a Jupyter Notebook, not sure if that affects anything.

Versions and main components

• PyMC3 Version: 3.6
• Theano Version: 1.0.4
• Python Version: 3.6.8
• Operating system: Arch Linux (exact build: 4.19.4-arch1-1-ARCH)
• How did you install PyMC3: (conda/pip) conda

[brandonwillard]

What's your version of Numpy (e.g. pip list | grep numpy)?

[MaxHalford]

It's 1.15.4.

[brandonwillard]

Hmm, the numpy.core.multiarray module is loaded as a linked library (shared object) on my system, so perhaps there were issues installing/building that on yours?

[MaxHalford]

Yeah I'm not sure what's going on... I just did conda create -n myenv python=3.6 pymc3, activated the myenv environment and the piece of code I wrote above still doesn't work.

[brandonwillard]

Conda might be reusing/sharing the same broken numpy library, so that every time you create a new env it's still the broken one.

What does import inspect, numpy; inspect.getfile(numpy.core.multiarray) give you?

[MaxHalford]

The command you gave me outputs:

'/home/max/anaconda3/lib/python3.6/site-packages/numpy/core/multiarray.py'

I assume it's using the broken numpy from my global installation and not installed it from scratch?

[brandonwillard]

Yeah, that look like a shared (and potentially broken) numpy. Try [re]installing numpy manually in one of those virtualenvs.

Also, this looks like a Theano-specific issue, so we have to move this to a Theano issue before going further.

Otherwise, if you don't care about C compilation, you might be able to disable it (for now) and use PyMC3. (Something like theano.config.cxx = '' might do the trick.)