meissnereric commented on issue #9295: test_operator.test_laop_3 hangs URL: https://github.com/apache/incubator-mxnet/issues/9295#issuecomment-357670401 We?ve been working on it this morning and have narrowed the issue down to a problem with syevd, only reproducible in the test environment. Running the following simplified code: ```python def test_laop_TEST(): np.random.seed(1896893920) n=10 data_in1 = np.random.normal(0., 10., (n, n)) data_in1 = 0.5 * (data_in1 + data_in1.T) u, lam = mx.nd.linalg.syevd(mx.nd.array(data_in1, dtype=np.float64)) print(data_in1, u, lam) assert(False) ``` Inside the test environment (using ?PYTHONPATH=./python/ nosetests3 --verbose tests/python/unittest/test_operator.py:test_laop_TEST?) returns the following results: A: [[-13.34414296 15.99829902 -5.9348693 0.33725903 7.08999108 3.48517748 -8.1782682 2.64821474 -11.71583022 -2.67093716] [ 15.99829902 -1.24079583 -5.15169143 -1.66193884 7.78811657 -3.24022297 6.49511542 12.28121039 8.40569059 -2.6123374 ] [ -5.9348693 -5.15169143 18.40724804 -3.32434622 -8.2294941 -12.05570085 -3.134901 -12.91228277 11.51627784 -3.26478123] [ 0.33725903 -1.66193884 -3.32434622 13.99603538 5.64011818 -5.48959896 4.85978757 4.35528089 -5.7480471 -11.40238071] [ 7.08999108 7.78811657 -8.2294941 5.64011818 -5.28942777 -3.71190675 -13.53220115 3.52255819 6.93703998 8.68193436] [ 3.48517748 -3.24022297 -12.05570085 -5.48959896 -3.71190675 -6.68378713 9.05448277 7.37555848 3.11379519 8.66276378] [ -8.1782682 6.49511542 -3.134901 4.85978757 -13.53220115 9.05448277 -1.94185915 0.70195063 -10.21748333 4.8669592 ] [ 2.64821474 12.28121039 -12.91228277 4.35528089 3.52255819 7.37555848 0.70195063 -20.63472654 7.87549062 1.3931764 ] [-11.71583022 8.40569059 11.51627784 -5.7480471 6.93703998 3.11379519 -10.21748333 7.87549062 3.42171743 4.7941094 ] [ -2.67093716 -2.6123374 -3.26478123 -11.40238071 8.68193436 8.66276378 4.8669592 1.3931764 4.7941094 -5.61335189]] U: [[-0.40939395 -0.02004096 0.27457066 -0.25153737 0.66140858 0.28340247 0.14125274 0.22853749 0.18226673 -0.26486544] [ 0.4472236 -0.48209223 0.10682268 -0.08815147 0.06254703 -0.36772867 0.30321108 0.56092533 0.04835745 -0.01863414] [-0.39439743 0.09942506 -0.01934432 -0.06989485 -0.3580831 0.07148518 -0.39629101 0.70887307 -0.11974635 0.14861161] [ 0.47108087 -0.15032447 0.34594981 0.19559081 -0.07895447 0.72793988 -0.23360136 0.07264473 -0.01704545 -0.03578629] [-0.07231265 0.06808902 0.2476193 0.26860241 0.20379301 -0.02531744 0.10901297 0.01760572 0.27485097 0.85250383] [-0.16324278 -0.30664205 -0.10891006 0.40596886 -0.10056171 -0.10892519 -0.28055657 -0.02900726 0.72918847 -0.26344591] [ 0.12151877 0.6399878 0.30715681 0.04750568 -0.33368478 -0.02954461 0.40925563 0.16125632 0.34502726 -0.23300369] [-0.34626999 -0.24576316 -0.15655482 0.51831127 -0.17288398 0.26124855 0.58672901 0.07084532 -0.27323606 -0.06688774] [ 0.24814453 0.40352612 -0.33373039 0.52155155 0.48083476 -0.10008073 -0.17301593 0.27878979 -0.1649485 -0.12694282] [-0.16370208 -0.05674461 0.6998121 0.32548404 0.00940855 -0.39724873 -0.21029293 -0.12432748 -0.34877216 -0.18731605]] <NDArray 10x10 @cpu(0)> Lambda: [-55.42942652 -36.08229736 -21.43322708 -13.37817004 -6.12139065 4.87496436 6.50358305 15.6977825 29.92340939 56.52168194] <NDArray 10 @cpu(0)> Whereas running the exact same code in an ipython notebook returns the following: A: [[-13.34414296 15.99829902 -5.9348693 0.33725903 7.08999108 3.48517748 -8.1782682 2.64821474 -11.71583022 -2.67093716] [ 15.99829902 -1.24079583 -5.15169143 -1.66193884 7.78811657 -3.24022297 6.49511542 12.28121039 8.40569059 -2.6123374 ] [ -5.9348693 -5.15169143 18.40724804 -3.32434622 -8.2294941 -12.05570085 -3.134901 -12.91228277 11.51627784 -3.26478123] [ 0.33725903 -1.66193884 -3.32434622 13.99603538 5.64011818 -5.48959896 4.85978757 4.35528089 -5.7480471 -11.40238071] [ 7.08999108 7.78811657 -8.2294941 5.64011818 -5.28942777 -3.71190675 -13.53220115 3.52255819 6.93703998 8.68193436] [ 3.48517748 -3.24022297 -12.05570085 -5.48959896 -3.71190675 -6.68378713 9.05448277 7.37555848 3.11379519 8.66276378] [ -8.1782682 6.49511542 -3.134901 4.85978757 -13.53220115 9.05448277 -1.94185915 0.70195063 -10.21748333 4.8669592 ] [ 2.64821474 12.28121039 -12.91228277 4.35528089 3.52255819 7.37555848 0.70195063 -20.63472654 7.87549062 1.3931764 ] [-11.71583022 8.40569059 11.51627784 -5.7480471 6.93703998 3.11379519 -10.21748333 7.87549062 3.42171743 4.7941094 ] [ -2.67093716 -2.6123374 -3.26478123 -11.40238071 8.68193436 8.66276378 4.8669592 1.3931764 4.7941094 -5.61335189]] U: [[ 0.56896406 -0.49508446 -0.05515309 -0.09274153 0.05952866 -0.30276783 0.41076905 0.18170427 0.34030827 -0.08221328] [-0.24222508 -0.14435087 0.26295741 -0.18396091 0.30094882 0.01134089 0.04484504 0.75461613 -0.35400702 -0.17327796] [-0.13275656 -0.00823993 0.00394372 -0.16026696 0.57617405 0.29618633 0.3182481 -0.42766506 0.05309422 -0.4999056 ] [ 0.4795233 -0.15113111 0.35068639 0.19762309 -0.06883189 0.72410954 -0.22482745 0.06161804 -0.02440456 -0.03418712] [ 0.1443947 0.08427882 0.32210052 0.15902208 0.42793043 -0.1065733 0.30022962 -0.18552473 -0.29677106 0.6602256 ] [ 0.4100674 0.61100202 0.24203264 -0.36583436 -0.26012926 -0.14006751 0.21300133 -0.01063569 -0.25163655 -0.26460707] [-0.26364054 0.32696231 0.18616994 0.38349845 -0.17037144 0.19078566 0.53068931 0.2664058 0.47268578 0.01995924] [ 0.19833204 0.25956163 0.15391535 0.52816029 0.37924181 -0.39832189 -0.39632359 0.06842959 0.15922644 -0.31789337] [ 0.06100089 0.28041907 0.00365086 -0.52407767 0.32170882 0.11788932 -0.29540979 0.16579616 0.55700809 0.31712067] [-0.2616713 -0.27894058 0.76428004 -0.18169327 -0.20501889 -0.23708203 -0.12629573 -0.27049638 0.21217396 -0.06485779]] <NDArray 10x10 @cpu(0)> Lambda: [-39.53770898 -31.85235173 -27.55026183 -13.38111156 -6.97443774 5.59771442 7.56903358 22.57472579 27.01218163 37.619126 ] <NDArray 10 @cpu(0)> In both cases, A is the same but U and Lambda are different. Thoughts?
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
