-
|
Hello Huanchen, I would like to use the Fiedler vector to reorder orbitals based on the mutual information matrix for spin-adapted calculations (SU(2)). However, I am getting errors when I try to run Is there a way to get the mutual information matrix based on the DMRG sites used in the SU(2) calculation (which I suppose would be a "CSF interaction matrix" as opposed to an "orbital interaction matrix")? Thanks, |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
|
The implementation of mutual information is based on the literature: J. Chem. Theory Comput. 9, 2959-2973 (2013). There is no mutual information matrix implementation in the SU(2) mode simply because I am not aware of any published research work describing how to do it. Of course you can translate the MPS from SU2 to SZ (https://block2.readthedocs.io/en/latest/tutorial/qc-hamiltonians.html#Change-from-SU2-MPS-to-SZ-MPS) and get the mutual information. And then you can use the mutual information for reordering in a subsequent SU2 calculation. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you! |
Beta Was this translation helpful? Give feedback.
The implementation of mutual information is based on the literature: J. Chem. Theory Comput. 9, 2959-2973 (2013).
There is no mutual information matrix implementation in the SU(2) mode simply because I am not aware of any published research work describing how to do it.
Of course you can translate the MPS from SU2 to SZ (https://block2.readthedocs.io/en/latest/tutorial/qc-hamiltonians.html#Change-from-SU2-MPS-to-SZ-MPS) and get the mutual information. And then you can use the mutual information for reordering in a subsequent SU2 calculation.