test_inference.py

import numpy as np
import random
import pytest

import inference
import network
import utils

# def make_random_array(length, limit_max=1000):
#     a = np.random.random_integers(1, limit_max, length)
#     #a = np.random.random((1, length))
#     return a

# def make_random_coincidences(size, length, limit_max=1000):    
#     c = make_random_array(size, limit_max)

#     for i in range(length - 1):
#         c = np.vstack((c, make_random_array(size, limit_max)))

#     return c

# def make_random_lambda(size, length):
#     c = [make_random_array(size, size)]

#     for i in range(length - 1):
#         c.append(make_random_array(size, size))
        
#     return c
    


# EPSILON = 0.00000001
        

# def test_dens_over_coinc():
#     length = 16
#     how_many_coinc = 200
#     c = make_random_coincidences(length, how_many_coinc)
#     i = make_random_array(length)

#     ## test for network.ENTRY
#     ##
#     y = inference.dens_over_coinc(c, i, network.ENTRY)
    
#     ## size of y equals the number of coinc
#     assert y.shape[0] == how_many_coinc

#     ## assert on each element of y
#     for j in range(how_many_coinc):
#         expected = np.linalg.norm(c[j,:] - i)
#         expected = np.exp(- np.power(expected, 2) / 1.0)
        
#         assert expected - y[j] < EPSILON

#     ## test for network.INTERMEDIATE and network.OUTPUT
#     ##
#     c = make_random_coincidences(length, how_many_coinc, length - 1)
#     i = make_random_lambda(length, how_many_coinc - 1)
#     y = inference.dens_over_coinc(c, i, network.INTERMEDIATE)
    
#     ## size of y equals the number of coinc
#     assert y.shape[0] == how_many_coinc

#     ## assert on each element of y
#     for j in range(how_many_coinc):
#         expected = np.array([])

#         for k in range(length):
#             expected = np.append(expected, i[k][c[j][k]])
                    
#         assert np.prod(expected) - y[j] < EPSILON