Currently the CI tests output is extremely verbose so it can take a bit of time to actually find where the test fails (if it fails). This verbosity should be reduced (e.g. by setting swpref.fid to 0).
In addition, the test also generate a large number of warnings. Some of these are flagged by SpinW because of certain issues with the calculations (such as non-positive definite Hamiltonians) and should actually be tested for and suppressed. Others are Matlab warnings and deprecation warnings (e.g. using spgr instead of sym) and should be fixed in the test code.
Also should look at some flaky tests:
test_cmplxBase_equivalent_with_tri (issue with indexing)
test_incommensurate (with mex=1 gets issue with indexing)
test_gtensor_incomm (with mex=1 sometimes gets 3e+248 near zone centre)
test_sw_with_nExt (with mex=1 sometimes gets 2e+248)
mducle
commented
1 year ago
Currently the CI tests output is extremely verbose so it can take a bit of time to actually find where the test fails (if it fails). This verbosity should be reduced (e.g. by setting
swpref.fidto0).In addition, the test also generate a large number of warnings. Some of these are flagged by SpinW because of certain issues with the calculations (such as non-positive definite Hamiltonians) and should actually be tested for and suppressed. Others are Matlab warnings and deprecation warnings (e.g. using
spgrinstead ofsym) and should be fixed in the test code.Also should look at some flaky tests:
test_cmplxBase_equivalent_with_tri(issue with indexing)test_incommensurate(withmex=1gets issue with indexing)test_gtensor_incomm(withmex=1sometimes gets3e+248near zone centre)test_sw_with_nExt(withmex=1sometimes gets2e+248)