Thank you very much for your answers Ian and Andrei.
The smearing (5K) improved the situation. The spin up is 123.7 and the spin
down 123.3, but there is 0.3 spin in the wrong place. The energy is nearly
the same, only 0.0005 eV lower, thus at least in energetic terms i do not
expect this to be a problem.
All the calculations were performed with SpinPolarized = T, but as you said
andrei, the system has a strong tendency to converge to a non-spin polarized
calculation. I still have some room to play with DM.InitSpin. Maybe it is
possible to obtain a different solution.
The problem with radicals that i was referring was that in some cases when
you do ``molecular calculations`` for a doublet radical you get a lot of
spin contamination, the <S2> can be as large as 1.5 instead of the correct
value 0.75. This is due to the presence of low lying excited states, for
example a quartet state.
Thus I was wondering if the problem that i have to obtain the spin resolved
calculation for the SWCNT+F hybrid is related to the presence of different
solutions.
It would be important to know the physical mining of this behavior. I guess
that is related to the too strong hybridization of the F2p orbitals that you
mention andrei, because in the case of hydrogen, there is no problem.
Many thanks for your help!
Pablo
----- Original Message -----
From: "Andrei Postnikov" <[EMAIL PROTECTED]>
To: <SIESTA-L@listserv.uam.es>
Sent: Friday, March 14, 2008 7:35 PM
Subject: Re: [SIESTA-L] mulliken Qtot
On Fri, 14 Mar 2008, Pablo Denis wrote:
| Dear Siesta users,
| I have a question about mulliken Qtot. I made
| this question about one year go but I still donĀ“t find a reasonable
| explanation.
|
| If I attach a fluorine atom to a SWCNT I got Qtot up 123.5 and Qtot
down
| 123.5. Why I have half beta spin?
Because you either did not declare a spin-resolved calculation
( SpinPolarized = T ),
or not initialized it with spin-up not equal to spin-down
(using, e.g., block DM.InitSpin ),
or you did all this but your calculation converged to a
non-spin-polarized solution (too large broadening, too agressive
mixing, too strong hybridization of F2p states with something else
due to a structure error, ...)
| I would expect 124 up and 123 down.
Initialize a spin-polarized calculation and look what you'll get
after one iteration (this might help to distinguish between
the above possibilities)
| Is this something similar to the spin
| contamination observed in the calculation of radicals with for example
B3LYP
| and pople`s basis sets?
What's the problem with radicals, they come out magnetic just fine...
(even if localization of magnetic density might be slightly
underestimated).
Good luck
Andrei
+-- Dr. Andrei Postnikov ---- Tel. +33-387315873 ----- mobile
+33-666784053 ---+
| Paul Verlaine University - Institute de Chimie, Physique et Mat\'eriaux,
|
| Laboratoire de Physique des Milieux Denses, 1 Bd Arago, F-57078 Metz,
France |
+-- [EMAIL PROTECTED] ------
http://www.home.uni-osnabrueck.de/apostnik/ --+
__________ NOD32 2948 (20080314) Information __________
This message was checked by NOD32 antivirus system.
http://www.eset.com