Dice
Dice copied to clipboard
sanshar
Question on fp-AFQMC
Hi!
I wonder whether the current implementation of fp-AFQMC does support open-shell case? Indeed, the all available examples are about the closed-shell case only:
https://github.com/ankit76/fp_afqmc/tree/main https://github.com/sanshar/Dice/tree/master/examples/DQMC/free_projection
However, the Python's scripts seem to be capable of handling UCCSD wavefunction ...
Thank you in advance!
With best regards, Victor
Yes it supports RHF, UHF and GHF. I will push an example soon.
Sandeep.
On Wed, May 24, 2023 at 7:47 AM vvp-nsk @.***> wrote:
Hi!
I wonder whether the current implementation of fp-AFQMC does support open-shell case? Indeed, the all available examples are about the closed-shell case only:
https://github.com/ankit76/fp_afqmc/tree/main https://github.com/sanshar/Dice/tree/master/examples/DQMC/free_projection
However, the Python's scripts seem to be capable of handling UCCSD wavefunction ...
Thank you in advance!
With best regards, Victor
— Reply to this email directly, view it on GitHub https://github.com/sanshar/Dice/issues/10, or unsubscribe https://github.com/notifications/unsubscribe-auth/AABVW4FIBNHYWFJVMY2R3MTXHYNP5ANCNFSM6AAAAAAYNQAV4E . You are receiving this because you are subscribed to this thread.Message ID: @.***>
I added an open shell example that uses uccsd here https://github.com/sanshar/Dice/tree/master/examples/DQMC/free_projection/feo
Hi again!
I appreciate you sharing an open-shell example. Everything works out. I am just puzzled about producing both 'uhf.txt' and 'rhf.txt' files. Do both files are really needed to proceed with fp-AFQMC on top of UCCSD ('right') and PHMSD ('left') trial wavefunctions?
Thank you!
With best regards, Victor
This is an internal technicality. Both 'rhf.txt' and 'uhf.txt' refer to the same set of rohf orbitals. Note that uccsd is performed on top of rohf in this example, not uhf.
Hi again!
Note that uccsd is performed on top of rohf in this example, not uhf.
That is quite an interesting point. Please correct me if I am wrong, but PySCF automatically performs conversion of ROHF object into UHF one, and then it performs regular UCCSD calculation. In this way, you just enforce that spatial parts for the beta and alpha orbitals are equal. In my eye, UCCSD on top of ROHF orbitals is a tricky way to preserve pure spin state and avoid any spin contamination, isn't It?
With best regards, Victor Vysotskiy
Hi Victor,
This was a clarification that a symmetry-broken uhf state is not used in this calculation, and not a comment about spin contamination in the initial state. Although, it is possible to check in practice if spin contamination is less severe when using a rohf reference as opposed to uhf in uccsd and subsequent fp-afqmc calculations.