Saptdft reuse jk for scf #3256
Saptdft reuse jk for scf #3256
Conversation
|
Is this ready for review? If so, please check "Ready for review" in the front page. If not, please mark this PR as a draft. |
Yes, it is ready for review. Thank you for letting me know that I needed to check that box. |
Co-authored-by: Jonathon Misiewicz <jpmisiewicz@vt.edu>
…into saptdft_reuse_jk_for_scf
| @@ -541,7 +549,7 @@ def run_sf_sapt(name, **kwargs): | |||
| core.IO.set_default_namespace('dimer') | |||
| data = {} | |||
|
|
|||
| if (core.get_global_option('SCF_TYPE') == 'DF'): | |||
| if (core.get_global_option('SCF_TYPE') == 'DISK_DF'): | |||
There was a problem hiding this comment.
Why is this and only this specific to DISK_DF?
There was a problem hiding this comment.
Good catch seeing this in another function. I have corrected this to match the other sections of the code.
Co-authored-by: Jonathon Misiewicz <jpmisiewicz@vt.edu>
…into saptdft_reuse_jk_for_scf
There was a problem hiding this comment.
Looks good! @Awallace3 can you look at my comment regarding the DF_INTS_IO option? (See https://github.com/psi4/psi4/blob/master/psi4/src/psi4/libfock/jk.cc#L239)
The best way to test this is to make sure that MemDF will still run when JK is reused.
| # We want to try to re-use itegrals for the dimer and monomer SCF's. | ||
| # If we are using Disk based DF (DISK_DF) then we can use the | ||
| # DF_INTS_IO option. MemDF does not know about this option but setting | ||
| # it will be harmless there. | ||
| core.set_global_option('DF_INTS_IO', 'SAVE') |
There was a problem hiding this comment.
Can you double check that setting this does not change the DF algorithm to DiskDF? In the options docs for DF_INTS_IO it says IO caching for CP corrections, etc. Changing this selects Disk_DF over Mem_DF. Note that setting this forces DiskDFJK when SCF_TYPE=DF.
(see https://github.com/psi4/psi4/blob/master/psi4/src/read_options.cc#L1674)
Thanks Andy! Your careful consideration helped remove some undesirable functionality of DF_INTS_IO when trying to prefer MemDF over DiskDF in the SAPT(DFT) code (when applicable). |
…into saptdft_reuse_jk_for_scf
| @@ -237,22 +237,16 @@ std::shared_ptr<JK> JK::build_JK(std::shared_ptr<BasisSet> primary, std::shared_ | |||
| Options& options, bool do_wK, size_t doubles) { | |||
| std::string jk_type = options.get_str("SCF_TYPE"); | |||
| if (jk_type == "DF") { | |||
| // logic for MemDFJK vs DiskDFJK | |||
| if (options["DF_INTS_IO"].has_changed()) { | |||
There was a problem hiding this comment.
For reference, the old behavior of touching DF_INTS_IO forces DISK_DF was a deliberate change made here
2b53daa that seems non-intuitive in retrospect.
Description
Previously, SAPT(DFT)
scf_helper()calls would re-compute integrals for SAPT0 dimer and monomer computations. Furthermore, the DFTscf_helper()calls would also recompute 3-index integrals by regenerating the JK-object. This PR enables developers to pass in a jk object to scf_helper to not re-compute 3-index DF integrals if desired. This change was largely drafted during a meeting between @cdsgroup with @konpat on Dec. 13th, 2024. Further tested in a 7 Feb 2025 meetup.closes #3173
User API & Changelog headlines
Dev notes & details
scf_helper()to prevent re-computing integrals for SCF.Questions
Checklist
Status