Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Restore approx 0 rounding to decomp0 method (#4862)
In #4656 we added rounding to the output of the decomp0 method to handle a case where differing FP precision on windows environments was causing an expected result in running two_qubit_cnot_decompose on np.eye(4) with numpy 1.19.x installed leading to a hard failure in the qasm tests. This seemed to reliably unblock testing and make unit tests work reliably. However, that original fix from #4656 was superseded by #4835 which was a fix for a more general issue with the reproducibility of the decompositions and reverted. Since #4835 has merged we've been seeing an uptick in the failure rate on the same unitary qasm test that #4656 fixed, so the change in #4835 was not actually sufficient for the windows case. This commit restores the fix from #4656 to unblock CI and fix the reproducability of the decompositions across systems. Fixes #4856
- Loading branch information