Phase of H_SOC elements
The entire H_SOC Hamiltonian is seemingly erroneously multiplied by i. At least the assignment of HSOI and HSOR is not consistent with the original publication Malmqvist et al. 2002 and my (repeated) derivation. I have also done numerical tests to check whether there is some "hidden" backcorrection. Found none, the results come out exactly how it is written in the code explicitly. More description is in the attached pdf. I also attach the python notebook mentioned in the pdf and input file for the test I performed.
I have also created a branch "SOC_issue" with the corrections that are necessary from my viewpoint. All tests with SOC after this correction fail.
Edited by Sergey Bokarev