The performances of two double hybrids, namely B2PLYP and PBE0-DH, are tested over the large GMNTK30 benchmark and compared with the results obtained with the related global hybrids, B3LYP and PBE0 with the aim of defining if there is still room for the development on nonparametrized functionals at DH level. Beyond the intrinsic interest in figures, these functionals' pairs are chosen as representative of the parametrized (B2PLYP/B3LYP) and parameter-free (PBE0-DH/PBE0) approaches to density functional theory. The obtained results show that the behavior of the double hybrids in general parallel the performances of the corresponding global hybrids, thus showing that either using a parametrized or using a nonparameterized approach to design new double hybrids, the performances are generally ameliorated with respect to the corresponding global hybrids. Nevertheless, the accuracy of B2PLYP is still higher than that of PBE0-DH, especially for thermochemistry. Albeit a link between performances and functional physics is difficult to extricate, it could be argued that this last result is not surprising since both B3LYP and B2PLYP are tuned on this last property.
No takes yet. Share an insight, caveat, or question.
Bousquet et al. (2013) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: