| 9:45:38 AM PST - Thu, Nov 15th 2012 |
|
Yes, it would still do it iteratively, but using only one iteration as the linear equation would be now uncoupled. Effectively, it diagonalizes it in one iteration. Hence, if that's what you see when you try it, you may have created a solution for yourself.
Bert
Quote:Ohlincha Nov 15th 1:38 amCheers Bert.
Let me know if/when it makes it into the development version and I'll be happy to test it against the published data set.
Hopefully such a switch will be enough to make it work - I tried editing (in NWChem 6.1) a number of files in src/property:
aoresponse_giao_rhs.F, giao_b1_movecs.F, giaofock.F, hnd_giaox.F, hnd_giaox_zora.F, int_giao_1ega.F
and changing to
xfac = 0.0d0
c if (use_theory.eq.'dft') xfac = bgj_kfac()
but then again I would claim to know what I'm doing and it obviously still iteratively solves the matrix rather than diagonalise it.
/Andy |