If I read the papers correctly, for B3LYP the functional woudl have the following form:
XC HFexch 0.05 slater 0.95 vwn_1_rpa 0.19 lyp 0.81 becke88 nonlocal 0.72
Explicitly not including the HFexch in the NMR response calculation would require some minor modifications to make it feasible.
Bert
Quote:Ohlincha Nov 9th 12:15 amDear all,
My, probably naive, question is this: is it possible to use NWChem in its current state for the Wilson-Amos-Handy method to calculating NMR shifts?
The approach is described in Chem. Phys. Letters 1999, 312, 475-484 and J. Chem. Phys, 2000, 113(8), 2983-2989 and 'excludes the exact exchange from the calculation of the first-order perturbed wave function so that the coupled-perturbed KS can be diagonalized and doesn't need to be iteratively solved' (paraphrasing Magyarfalvi and Pulay's description on page 1351 in J. Chem. Phys. 2003, 119, 1350-1356).
At a first glance it may look like the method only involves a slight re-parametrisation (xc HFexch 0.05 slater 0.95 becke88 0.72 perdew 91 wvn_5 1), but the functional alone performs as poorly as b3lyp, acm etc. and there are more significant changes in the approach.
To my knowledge this has only been implemented in PQS so far.
I'm interested in the method given the reported excellent accuracy in the prediction of the 17O NMR shifts ('simple' test cases like CO and H2CO are not handled well by giao/b3lyp) for at least first and second row elements.
Is it possible to reproduce the Wilson-Amos-Handy method in the current version of NWChem, or would it involve non-trivial changes/additions to the code?
Best regards,
Andy |