Hi Adam, Thanks for your suggestion. It could be even simpler: as the electronic dipole moment alone is not origin-independent, I probably have to add terms with the overlap matrix times r with respect to the common origin to the dipole matrices.
Heribert On Tue, 2009-12-01 at 15:23 +0100, [email protected] wrote: > Hi Heribert, > > sorry, I thought of the transition dipole moment. I guess, it is important > to check how SIESTA tranforms the grid internally. Usually, the FFT is > faster if the origo is at the edge of the parallelopipedon (I guess, > it is simply a cube in your case) but you assume that the > origo is in the middle of the cube. Does it resolve your problem? > > Yours, > Adam > > > Thanks for your reply, however, I just need the ground state dipole > > moment, and a point of the project is that supercells are avoided, so I > > suppose a supercell calculation is not an option. Are you trying to say > > that the 'direct' calculation of dipoles (or transition dipoles) using > > the wavefunction in Siesta is not advisable? > > > > Heribert > > > > > > On Tue, 2009-12-01 at 13:45 +0100, [email protected] wrote: > >> Hi Heribert, > >> > >> I believe that it is more safe to calculate the dipole transition moment > >> in the velocity representation in a supercell code. For that you > >> need the derivative of the wave functions. > >> > >> Yours, > >> Adam > > > > > > ---------------------------------------------------------------------------- > Dr. Gali Ádám Adam Gali, PhD > > Budapesti Műszaki és Department of Atomic Physics, > Gazdaságtudományi Egyetem, Budapest University of Technology > Atomfizika Tanszék and Economics > Budapest, Budafoki út 8., 1111 Budafoki út 8., H-1111, Budapest, > Hungary > > telefon: 463-1580 telephone: [36]-(1)-463-1580 > fax: 463-4357 fax: [36]-(1)-463-4357 > > e-mail: [email protected] > http://www.fat.bme.hu/kulonc/galia/index.en.phtml > ----------------------------------------------------------------------------
