Mismatch in schur_complex_2 'schur<MatrixXcf>' when input size=2 is generated via internal::random<int>
Summary
After compilation of schur_complex.cpp and running schur_complex_2, mismatch is observed with 'schur' when input size=2 is generated from internal::random.
Environment
- Operating System : Linux
- Architecture : x64
- Eigen Version : 3.4.0
- Compiler Version : v6.3.0 g++
- Compile Flags : -O0 -g -m32 -pipe
- Vector Extension : SSE/AVX/NEON ...
Minimal Example
Running schur<MatrixXcf>(2)
i.e. CALL_SUBTEST_2(( schur<MatrixXcf>(2) ));
Steps to reproduce
- $ cd /tmp/
- $ unzip eigen-3.4.0.zip
- $ mkdir build && cd build
- $ <path_to/cmake/v3.17.3/bin/cmake -D EIGEN_TEST_CXX11=ON -D CMAKE_BUILD_TYPE=DEBUG -D CMAKE_C_COMPILER=<path_to_/gcc/v6.3.0/bin/gcc> -D CMAKE_CXX_COMPILER=<path_to/gcc/v6.3.0/bin/g++> -D EIGEN_SPLIT_LARGE_TESTS=ON -D EIGEN_TEST_MAX_SIZE=8 -D DART_TESTING_TIMEOUT=240 -D CMAKE_SYSTEM_NAME=Linux -D CMAKE_CROSSCOMPILING_EMULATOR= -D 'CMAKE_C_FLAGS= -O0 -g -m32 -pipe' -D 'CMAKE_CXX_FLAGS= -O0 -g -m32 -pipe' -D CMAKE_EXE_LINKER_FLAGS=-lm ../eigen-3.4.0
- $ make VERBOSE=1 -Bj schur_complex_2/fast
- $ ctest -j1 --timeout 100 --no-label-summary --output-on-failure --stop-on-failure --progress -V -R schur_complex_2
What is the current bug behavior?
Mis-match is observed at schur_complex.cpp (57)
What is the expected correct behavior?
There should not be any mis-match
Relevant logs
Initializing random number generator with seed 1647003388
Repeating each test 10 times
actual = 0
expected = 2
Test ( schur<MatrixXcf>(internal::random<int>(1,8/4)) ) failed in /tmp/eigen-3.4.0/test/schur_complex.cpp (57)
test_is_equal(cs3.info(), size > 1 ? NoConvergence : Success, true)
Stack:
- ( schur<MatrixXcf>(internal::random<int>(1,8/4)) )
- schur_complex
- Seed: 1647003388
0% tests passed, 1 tests failed out of 1
Total Test time (real) = 0.05 sec
The following tests FAILED:
447 - schur_complex_2 (Child aborted)
Errors while running CTest
----------------------------------------------------
----------------------------------------------------
-
Have a plan to fix this issue.
Edited by Rohan Ghige