[Numpy-discussion] Failed numpy.test() with numpy-1.10.1 on RHEL 6

Eric Moore

2015-11-09 13:24:51 UTC

This fails because numpy uses the function `cacosh` from the libm and on
RHEL6 this function has a bug. As long as you don't care about getting the
sign right at the branch cut in this function, then it's harmless. If you
do care, the easiest solution will be to install something like anaconda
that does not link against the relatively old libm that RHEL6 ships.

Post by Lintula
Hello,
I'm setting up numpy 1.10.1 on RHEL6 (python 2.6.6, atlas-3.8.4,
lapack-3.2.1, gcc-4.4.7), and this test fails for me. I notice that
someone else has had the same at
https://github.com/numpy/numpy/issues/6063 in July.
Is this harmless or is it of concern?
======================================================================
FAIL: test_umath.TestComplexFunctions.test_branch_cuts(<ufunc
'arccosh'>, [-1, 0.5], [1j, 1j], 1, -1, True)
----------------------------------------------------------------------
File "/usr/lib/python2.6/site-packages/nose/case.py", line 182, in runTest
self.test(*self.arg)
File
"/usr/lib64/python2.6/site-packages/numpy/core/tests/test_umath.py",
line 1748, in _check_branch_cut
assert_(np.all(np.absolute(y0.imag - yp.imag) < atol), (y0, yp))
File "/usr/lib64/python2.6/site-packages/numpy/testing/utils.py", line
53, in assert_
raise AssertionError(smsg)
AssertionError: (array([ 0.00000000e+00+3.14159265j,
1.11022302e-16-1.04719755j]), array([ 4.71216091e-07+3.14159218j,
1.28119737e-13+1.04719755j]))
----------------------------------------------------------------------
Ran 5955 tests in 64.284s
FAILED (KNOWNFAIL=3, SKIP=2, failures=1)
<nose.result.TextTestResult run=5955 errors=0 failures=1>
_______________________________________________
NumPy-Discussion mailing list
https://mail.scipy.org/mailman/listinfo/numpy-discussion