Posted by Grigory Shamov on February 18, 2010 at 21:37:00:
In Reply to: Re: Density Functionals, questions posted by Roland Lindh on February 17, 2010 at 08:45:23:
Dear Prof Lindh,
Thank you for your explanation!
: 1. This could be implemented but is not at this time. Such gradients would be possible to but would also require some new code since the wave functions is not variational.
There seems to be uses for post-SCF DFT computations -- namely, HF+WL (Wilson-Levy) functional for weak interactons works better post-SCF that self-consistently; some people have recently claimed that for double-hybrids it is better to have SCF done in something like B3LYP and then use the density for some HF+LYP+PT2 combination.
Even the energy only computations are interesting.
: 3. Just email me the stuff together with some documentation of what it is, please. I'll make sure that it is implemented in the development version.
Thank you!
Btw, what level of derivatives is needed for the present code? Are the second derivatives used for something/anything now? (So far I've checked the energies and first derivatives by comparing the total energies and gradients against published values, for small molecules).
--
With best regards,
Grigory Shamov
Chemistry,
University of Manitoba