Open jtkrogel opened 6 years ago
I think this is covered by diamond 2x1x1. However if there is some part of the tiling machinery we are not covering, the test should be expanded.
Not sure. My naive impression is that tiling by an integer number in a single direction is easier to get right (even in the face of errors) than tiling by an arbitrary matrix.
There is a reasonable candidate for this: a 4 atom carbon diamond cell (cubic) tiled from a 2 atom fcc cell. Luke created an example for this in 2013 and it could be imported directly as a test.
Reference data exists to 1 mHa (VMC/DMC) and the wavefunction file is only 880K.
Would it be possible to unit test the tiling? i.e. check that the correct consistent values are obtained at different equivalent (tiled) points in space?
Since we have the diamond 2x1x1, we might need just a unit test.