9
$\begingroup$

Hybrid DFT methods, where the functional is supplemented with Hartree-Fock exchange, have become increasingly popular due to their low cost and decent accuracy. Double hybrids, which mix in an MP2 contribution, have also been developed as a way to improve the accuracy.

Has this idea been extended to other Post-SCF methods? For example, one could consider doing coupled cluster or configuration interaction calculations where a DFT wavefunction is used as the primary configuration. If not, is there a major hurdle, technical or theoretical, that prevents this from being done?

$\endgroup$
2
  • 1
    $\begingroup$ @NikeDattani I had initially thought of asking about a "reverse hybrid", but after reading a bit more on double hybrids, apparently there isn't much of a distinction. Double hybrid DFT methods actually perform an MP2 calculation with the DFT wavefunction, so its not necessarily a clear distinction whether you are improving DFT with MP2/Post-SCF or the other way around. $\endgroup$
    – Tyberius
    May 16 '20 at 17:19
  • $\begingroup$ I once asked about mixing in coupled cluster. The main reluctance to mix in CC is probably that the cost would become $N^6$ for CCSD, which might defeat the purpose of DFT (some people might say that the purpose of DFT is to provide a low-cost method). The nice thing about DFT+MP2 is that the accuracy tends to be better than standard MP2, and gets quite close to CCSD(T) accuracy with roughly MP2 cost. But let me ask Stefan Grimme for a better answer! $\endgroup$ May 30 '20 at 4:45
4
$\begingroup$

As far as I am aware, double hybrids were first proposed by Truhlar and coworkers in J. Phys. Chem. A 2004, 108, 21, 4786–4791. If you look at the paper, in addition to MP2 they also proposed a double hybrid based on QCISD, i.e. the quadratic configuration interaction method which can be viewed as an approximate coupled-cluster method.

There's no reason a priori why more elaborate post-HF methods than MP2 couldn't be employed in a double hybrid DFT functional. However, the reason why this isn't usually done is probably due to cost and accuracy: going to higher-order methods introduces much steeper scaling with system size (already MP2 is quite expensive!), while the improvements in accuracy are probably quite low and still dominated by the DFT part: at some point it's just better to discard DFT altogether and just use the post-HF method!

Note also that in addition to MP2, some double hybrids employ the random phase approximation, see e.g. Phys. Chem. Chem. Phys. 18, 20926 (2016)

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.