Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly  
We are NATO-free zone

Re^3: Increasing weight of exact HF-exchange in functional (option HFX) in TDDFT calculations

Thomas Pijper

Dear Evgeniy,

Such a modification is allowed. The point I was trying to make is that BH&HLYP can actually be considered as a version of B3LYP with HFX set to 0.5 (the only difference being that B3LYP uses a small 'excess' of local exchange and local correlation while BH&HLYP doesn't).

B3LYP exchange: B88 (0.72) + Slater (0.08) + exact (0.2)
B3LYP correlation: LYP (0.81) + VWN (0.19)

BH&HLYP exchange: B88 (0.5) + exact (0.5)
BH&HLYP correlation: LYP (1.0)

As such, for B3LYP I do not see a problem with setting a different value for HFX.

Kind regards,

On Fri Aug 5 '16 6:06pm, GrEv wrote
>Dear Thom,

>Thanks very much for your reply. Indeed, the BHHLYP functional has been used a lot in TDDFT calculations. Is seems to make the charge-transfer states higher in energy indeed. In fact, it makes overestimate of such states in energy quite much, at least in my case. On the other hand, if I increase the weight of HF-exchange in the B3LYP functional from 0.2 to 0.3 (using the HFX option in Firefly) I reproduce the results for the charge-transfer states obtained with an ab-initio (CC2) method. The question is/was if such a kind of modification of B3LYP is "allowed" and might have been used before?

>Best regards,
>On Wed Aug 3 '16 9:28pm, Thomas Pijper wrote
>>Dear Evgeniy,

>>This is indeed possible. The BH&HLYP functional (which uses 50 % exact exchange) is actually a commonly used functional for TDDFT calculations. References that use this functional should be plentiful.
>>Kind regards,
>>On Wed Aug 3 '16 1:44pm, GrEv wrote

>>>I wonder if one can use the option HFX, to increase the weight of exact HF-exchange in a hybrid functional (B3LYP), to "remedy" the problem with charge-transfer states in TDDFT calculations. I would appreciate very much a reference to some works where this was used.

>>>Best regards,

[ Previous ] [ Next ] [ Index ]           Fri Aug 5 '16 9:23pm
[ Reply ] [ Edit ] [ Delete ]           This message read 371 times