BeliefPropagation fails with IndexError

Subject of the issue

BeliefPropagation fails with error IndexError: only integers, slices (":"), ellipsis ("..."), numpy.newaxis ("None") and integer or boolean arrays are valid indices.

Your environment

• pgmpy version: 0.1.15
• Python version: Python 3.7.12
• Operating System: Google Colab

Steps to reproduce

import os
from pgmpy.readwrite import BIFReader
def load_network(network_name):
  url = f"https://www.bnlearn.com/bnrepository/{network_name}/{network_name}.bif.gz"
  os.system(f"wget {url} -q")
  fn = f"{network_name}.bif.gz"
  os.system(f"gzip -qd -f {fn} -q")
  fn = f"{network_name}.bif"
  reader = BIFReader(fn)
  os.system(f"rm {fn}")
  model = reader.get_model()
  model.states = reader.get_states()
  
  return model

network_name = 'asia' #@param ["asia", "cancer", "earthquake", "sachs", "survey"]
model = load_network('asia')

from pgmpy.inference import BeliefPropagation
belief_propagation = BeliefPropagation(model)
belief_propagation.query(variables=['lung', 'bronc'],
                                         evidence={'asia': 'yes', 
                                                           'tub': 'no', 
                                                           'xray':'yes'})

This produces IndexError: only integers, slices (":"), ellipsis ("..."), numpy.newaxis ("None") and integer or boolean arrays are valid indices.

[ankurankan]

Interesting. This throws the error for me only sometimes and works fine other times. I will check why this is happening. Thanks again for reporting this.

[ankurankan]

Also, just FYI an easier way to load models from the bn repository is to use get_example_model:

from pgmpy.utils import get_example_model
asia = get_example_model('asia')