Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

test_interpolator fails with quite a swing: -3.4 vs 3

See original GitHub issue

smaug:/tmp/buildd/nipy-0.4.3~rc1# python3 -m nose .pybuild/cpython3_3.9_nipy/build/nipy/algorithms/tests/test_interpolator.py:test_interpolator
nipy.algorithms.tests.test_interpolator.test_interpolator ... FAIL

======================================================================
FAIL: nipy.algorithms.tests.test_interpolator.test_interpolator
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/lib/python3/dist-packages/nose/case.py", line 197, in runTest
    self.test(*self.arg)
  File "/build/nipy-0.4.3~rc1/.pybuild/cpython3_3.9_nipy/build/nipy/algorithms/tests/test_interpolator.py", line 23, in test_interpolator
    assert_almost_equal(interp.evaluate([0, 0, 4]), arr[0, 0, -1])
  File "/usr/lib/python3/dist-packages/numpy/testing/_private/utils.py", line 579, in assert_almost_equal
    return assert_array_almost_equal(actual, desired, decimal, err_msg)
  File "/usr/lib/python3/dist-packages/numpy/testing/_private/utils.py", line 1042, in assert_array_almost_equal
    assert_array_compare(compare, x, y, err_msg=err_msg, verbose=verbose,
  File "/usr/lib/python3/dist-packages/numpy/testing/_private/utils.py", line 840, in assert_array_compare
    raise AssertionError(msg)
AssertionError: 
Arrays are not almost equal to 7 decimals

Mismatched elements: 1 / 1 (100%)
Max absolute difference: 6.4
Max relative difference: 2.13333333
 x: array(-3.4)
 y: array(3)
>>  raise AssertionError('\nArrays are not almost equal to 7 decimals\n\nMismatched elements: 1 / 1 (100%)\nMax absolute difference: 6.4\nMax relative difference: 2.13333333\n x: array(-3.4)\n y: array(3)')

Issue Analytics

State:
Created 3 years ago
Comments:6 (6 by maintainers)

Top GitHub Comments

3reactions

matthew-brettcommented, Feb 22, 2021

The key difference is here:

# Test 2D interpolation for Scipy
import numpy as np

import scipy.ndimage as spnd

arr = np.arange(6).reshape((2, 3))

coord = np.array([[0, 3]]).T
v1 = spnd.map_coordinates(arr,
                          coord,
                          order=3,
                          mode='nearest',
                         )
print('Calculate filter with interpolation', v1)

filtered = spnd.spline_filter(arr, order=3, mode='nearest')
v2 =  spnd.map_coordinates(filtered,
                           coord,
                           order=3,
                           mode='nearest',
                           prefilter=False)
print('Calculate filter followed by interpolation', v2)

On Scipy 1.2 this gives [2.] and [2.]. On Scipy 1.6.6 this gives [2.] and [2.20261874]. So it’s the difference between using precalculated knots via spline_filter, or calculating the knots afresh within map_coordinates.

I’ll file an issue with Scipy, and see what they say.

0reactions

matthew-brettcommented, Mar 29, 2021

I think this is fixed by #478 and #481 .