librascal
librascal copied to clipboard
cutoff in tests too small
In test_adaptor_strict.cc
is a multiple_fixtures
list with
MultipleStructureFixture<MultipleStructureManagerNLFixture>,
MultipleStructureFixture<MultipleStructureManagerNLCCFixture>
We just noticed that at least parts of the underlying fixtures which are in there have a cutoff of 1, which is too small for the given structure. I.e. the neighbourlist does not contain any pairs. The respective structures should be changed to an appropriate cutoff. Otherwise nothing is tested here.
@Felix: small cutoff is intentional to check for empty neighbourlist
test_adaptor.hh MultipleStructureManagerNLFixture
:
Try do reduce the number of cutoff to two, which are intentional to check for neighbours, etc.
It would be helpful for the fixtures to have a member stating whether the adaptor-cutoff combination should lead to a managers for which some centres have no neighbour.