FEA-Bench / testbed /fairlearn__fairlearn /test /unit /utility_functions.py
hc99's picture
Add files using upload-large-folder tool
fc0f7bd verified
# Copyright (c) Microsoft Corporation. All rights reserved.
# Licensed under the MIT License.
import numpy as np
def logging_all_close(a, b, rtol=1e-05, atol=1e-08, equal_nan=False):
"""Wrap numpy.all_close functionality with print() on failure.
Function arguments match numpy.all_close (assuming that it uses
numpy.isclose), but if there are failures, then some diagnostics
will be printed to stdout.
"""
aa = np.asarray(a)
ba = np.asarray(b)
match_mask = np.isclose(aa, ba, rtol=rtol, atol=atol, equal_nan=equal_nan)
if not np.all(match_mask):
print("a mismatches: ", aa[np.logical_not(match_mask)])
print("b mismatches: ", ba[np.logical_not(match_mask)])
print("mismatch indices: ", np.where(np.logical_not(match_mask)))
return np.all(match_mask)