Even though I'm not a communist ...
For a metal even such Pulay can be to big ... I have cases when I got
convergence only with 0.005 ...
Andrey's typs are also very helpful
2010/3/28 guohuaihong
> Dear SIESTA-communist,
>
> I am a SIESTA green hand, and I try to get self-consistent calculation
Dear developer,
I am a graduate student in Nanjing University of Technology. I am very
interested in you wonderful code-Siesta.
Could you please give us a license to acess your code.
I filled the form in your web page to apply for the license a couple
days ago. But our application is still in
nicolags,
it could be that your system is fluctuatin between semi-metallic and
semi-conducting phases in the scf interations...you may try many things...
1. increase # of k-points
2. decrease the mixing
3. try to increase broadening in dos...
s.auluck
> Hello S. Auluck,
>
> Thank you for your an
Hello S. Auluck,
Thank you for your answer. But as mentioned to Alberto, my system is still
semi-metallic (bandgap crossing not far from K k-point), so no bandgap is
present... I will do some tests on pure graphene to see if the problem
occurs there too.
Nicolas
On Sun, Mar 28, 2010 at 9:50 AM,
> Dear SIESTA-communist,
I wonder how many in the community will feel addressed...
Anyway, here is my guess:
You set spin = true but do not define DM.InitSpin block
(unless I overlooked it). In this case all atoms,
including Ga, are initialized with maximal spin moment.
Then the system tries to re
hi,
in a semi-conductor or insulator there is a energy gap. in this region the
dos is zero. so in principle E_f can be placed anywhere in between because
E_f is determined by the integral of N(E)dE from zero to E_f which must
give the total number of valence electrons.
it depends on the condition