New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Renaming of excited states dipoles #2541
Conversation
I hope you know what you're asking for. Storage of ground-state densities is lawless, let alone excited-state densities. For excited-state densities...
For ground-state densities...
And to represent all that, the current tools we have are I'm in favor of deprecating |
Thanks for the enumeration of the organizational state of all the densities. The new Wfn member for finalized, labeled densities sounds good to me. Are the ES CC densities ready for that state yet, or are they still in preparation? For any that are still in preparation but still in need of saving, I guess I propose a temporary member of Wfn that is marked as having a finite lifetime. Same as Wfn.arrays, just not Wfn.arrays :-) Ben's already nervous about storing the per-atom charges that end up in qcvars, so I'm just not in favor of densities there. |
The CC excited state densities are most definitely correct. If they weren't, EOM analytic gradients would be wrong, and I don't know whether the transition densities pass some kind of finite difference test, but I am prepared to save that question for another day. |
Can you clarify what you mean about a "finite difference test" for the transition densities? |
By finite-difference test, I mean "there is some property that we can compute either by finite difference of energies or by contracting appropriately defined densities against derivative integrals," e.g., geometry gradients and dipoles. By checking that both routes predict the same result, we can be much more confident that the densities are correctly implemented. (I recall a sign error in the CASPT2 gradients of another package. This error went uncaught for decades because the impact on calculations was relatively small.) I don't know if this is an option for transition densities. I'm not sure if EOM-CC transition properties are defined by some variational criteria, some variational criteria but neglecting orbital relaxation, or something else altogether. EDIT: The '93 Stanton and Bartlett paper explicitly says orbital relaxation is neglected, so I imagine the finite difference test is not an option. I'm not sure if there's some other technique to validate the correctness of the transition densities, other than matching other code. I know "matching other code" is done in the test suite. |
OK, then we're on the same page. Transition properties aren't clearly defined as derivatives, so I don't see how a finite-difference approach can help in this case, but we might be able to identify some limiting cases. But I agree it can/should wait for another day. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
apart from lodging the root densities outside Wfn.arrays, lgtm!
Co-authored-by: Lori A. Burns <lori.burns@gmail.com>
Okay, latest version of the code moves the excited state densities to a new field. Unless anybody has grievances, we'll start trying this out. |
/// Vector of alpha density matrices | ||
std::map<std::string, SharedMatrix> Da_map_; | ||
/// Vector of beta density matrices | ||
std::map<std::string, SharedMatrix> Db_map_; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
if Da_map_
and Db_map_
become the long-term holders, should probably add some of the key.upper()
features of variables and arrays, but that's for the beyond experimental stage.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm not really qualified to appraise this, but I don't see any obvious errors.
Alas, nobody is fully qualified to appraise this. TDC knows the I won't be free to work on the next PR for a while, so we can leave this open to collect more comments. |
Description
More
cc
tests ported over. There's oneautotest
test left.Todos
Checklist
Status