For an AFM material, where two types are needed for the same element, I labeled them 'Mnu' and 'Mnd' for Manganese up and down.
SelfConsistentHubbardWorkChain let me go ahead and completed one entire cycle (relax – scf’s – hp). However, it crashed before the first relaxation of the second cycle because it needed names to be 'Mn1' and 'Mn2' (not 'u' and 'd'). I understand the relabeling is part of a broader issue of book-keeping, so that’s fine. However, it would be nice if the WC rejected ‘Mnu’ and ‘Mnd’ from the beginning, rather than letting the user complete one entire cycle knowing it will crash at the start of the second cycle (in cases of builder.meta_convergence = Bool(True) anyway).
This affects branch new_outline+spin_fix. Please feel free to close the issue if it's already sorted in branch merge/scfwc/U/U+V.
For an AFM material, where two types are needed for the same element, I labeled them 'Mnu' and 'Mnd' for Manganese up and down.
SelfConsistentHubbardWorkChainlet me go ahead and completed one entire cycle (relax–scf’s –hp). However, it crashed before the first relaxation of the second cycle because it needed names to be 'Mn1' and 'Mn2' (not 'u' and 'd'). I understand the relabeling is part of a broader issue of book-keeping, so that’s fine. However, it would be nice if theWCrejected ‘Mnu’ and ‘Mnd’ from the beginning, rather than letting the user complete one entire cycle knowing it will crash at the start of the second cycle (in cases ofbuilder.meta_convergence = Bool(True)anyway).This affects branch
new_outline+spin_fix. Please feel free to close the issue if it's already sorted in branchmerge/scfwc/U/U+V.