Carlos Reis wrote:
Hello Since there is some discussion about spin polarized calculations I shall repost my questions that went unanswered: I need some clarification on the options DM.InitSpin and DM.InitSpinAF: 1. How does Siesta defines/imposes the spin state/configuration of an atom in a polyatomic system?
In the first SCF step, the atomic pseudo-orbitals are populated with electrons with spin up/down depending on the configuration chosen. In the AF option, all the electrons are up in an atom, and down in the nearest-neighbours, and so on.
2. After Self Consistency is the spin of that atom conserved? if so How ?
No. Sometimes, you start from a particular spin configuration that is not stable, and end up in a different magnetic state.
3. What about de-localized (and occupied) states? 3a Don't we have to integrate the spin density in some region about a certain atom in order to know the spin of of that atom?
You are right. You have to choose a region of integration. Notice that SIESTA only populates atomic orbitals with spins in the first SCF step (when no data is available). The final SCF solution is a combination of atomic orbitals, and the spins are distributed in a "non-local" way. The definition of "atomic magnetization" is not clear. What you can do is choose a particular definition (for example, based in Mulliken population). This is a random definition. You can plot the electronic charge or magnetic distribution and define your own "atomic volume".
4. Does Siesta constraints the spin of an atom during SCF? if so How?
I don't think the "official" version has this option, but there might be user-modifications to that... Hope that helps. Regards, Miguel -- Miguel Pruneda Work Phone: +44 1223 333409 Dept. of Earth Sciences Home Phone: +44 1223 708849 Downing Street, CB2 3EQ Fax: +44 1223 333450 Cambridge, U.K. Email: [EMAIL PROTECTED]
