> I should have been more specific. Let's say I simulate one hydrogen atom > that has 1 electron in its valence. A non-spin-polarized calculation runs > successfully. I have read that the "extra" odd electron can be treated as > having half of its regular spin value in the spin up state and another > half in the spin down state. I suppose this is what siesta must do. > Chris
Dear Chris, yes, in SIESTA (as in any other code programmed without mistakes), for the case rho(up)=rho(down), the exch-corr potential V_ex(up)=V_ex(dn), if this answers your question... Hence the band structure will be the same (only the effort doubled) as if calculated without spin polarization. This option (no spin polarization, spin up = spin down) occurs by default if you set spin polarization to false. The code does not care about your number of electrons. However, YOU should care about what you are doing. Notably, a solution for hydrogen atom with a half-electron spin up and half spin-down is physically very wrong (but, of course, makes a perfectly legal test case). Very briefly, the options to consider with respect to your initial question (odd number of electrons) are: i) your system is METAL, you know that it is nonmagnetic (like Na); you do not need to switch on spin polarization... but take care of k-point conergence, check that your density of states is reasonable; ii) your system is a MOLECULE, or an INSULATOR; then, since number of spin up not equals spin down, you must include spin polarization, and initialize the magnetic solution somehow (setting initial magnetic moment on some atoms). Otherwise you may end up with a situation where topmost single level is occupied by a half-electron, which is neither physically reasonable nor numerically stable. Best regards Andrei Postnikov > 2010/9/13 leila <[email protected]> > >> Dear Akbar, >> >> >> >> Set SpinPolarized to be true in the input file. The total energy is >> calculated on both spin_up and _down electrons density, as we finally >> obtained in *.RHO file. >> >> If my opinion is not correct, I am glad if you point it out. >> >> >> >> Best wishes >> >> Yours sincerely, >> >> Leila >> >> *������:* akbar jahangiri [mailto:[email protected]] >> *����ʱ��:* 2010��9��14�� 11:39 >> *�ռ���:* [email protected] >> *����:* Re: ��: [SIESTA-L] odd number of electrons and not spin-polarized >> >> >> >> Dear Leila >> >> How to add spin in our calculation(when it is not spin polarised)? what >> do >> you mean by that? >> >> Best regards >> Akbar >> >> 2010/9/14 leila <[email protected]> >> >> Dear Chris, >> >> I just give my opinion. >> >> Just add spin in your calculation, if the electrons do not contribute >> to >> the magnetic moment, the additive component to the Hamilton only >> influences >> the absolute energy but not the relative energy. >> >> >> >> Best wishes! >> >> Yours sincerely, >> >> Leila >> >> *������:* Christopher Rowan [mailto:[email protected]] >> *����ʱ��:* 2010��9��14�� 6:23 >> *�ռ���:* [email protected] >> *����:* [SIESTA-L] odd number of electrons and not spin-polarized >> >> >> >> Hello, >> I'm wondering how SIESTA deals with a system that is not spin-polarized >> and >> yet has an odd number of electrons? >> Can anyone clarify? >> Chris >> >> >> >
