Incorrect orbits determined for BCC lattice about kgridgen

[glwhart]

How do you know that they are incorrect? And if so, what's going wrong, do you think?

[KaceyLeavitt]

The code placed the k-point [1/9, -5/9, -1/9] in an orbit with weight 1 when it should also have the k-point [-1/9, 13/9, 1/9] in its orbit. The code also placed the k-point [-.5, -.5, .5] in an orbital of weight two, but that k-point should be in an orbital of weight 1. I think this problem stems from an issue with indices in the hashtable.

[KaceyLeavitt]

By making a slight change to the hashtable indexing, the issue was fixed and the unit test is now passing.

[glwhart]

Did you commit the change to master?

Can you be specific about what was wrong with the indexing and how you changed it? We need to be more verbose...

[jerjorg]

It was committed in revision 0.2.6. Here is the line of code that changed

! iFirst(cOrbit) = idx
iFirst(cOrbit) = iUnRdKpt

Instead of having iFirst point from cOrbit to the index of a k-point that represented the orbit, it was changed so that cOrbit pointed to the position of the representative k-point in UnreducedKpList.

[glwhart]

Great. I see what the problem was now. Thanks