[Smeagol-discuss] problem regarding convergence for antiparallel spin confi

[Ivan Rungger]

Dear Kanchan,

  for anti-parallel calculations you have to set the initial spin by 
hand in the initspin block, + for the left Ni atoms and - for the right 
Ni atoms (assuming the left leads is up and the right lead is down). For 
the other molecules you don't have to initialize the spin. It is very 
likely that if you don't do that the calculations does not converge 
(even if it would converge the result would not be correct if you don't 
initialize  correctly the Ni electrodes up and down).

Cheers,

Ivan

ulman at jncasr.ac.in wrote:
>
> Dear All,
>
> I have done the calculations for Nickel leads with parallel spin
> configuration and these calculations have converged well. For the
> calculations with antiparallel spin configuration of the leads, I started
> the transport calculations using the spin up leads files for left lead and
> spindown leads for the right lead. My calculations donot seem to converge
> at all. I tried playing around with the DMmixingweight but even that
> doesnot help. The values of dHMmax go down to 6*10^-3 and then starts
> increasing steadily. Sometimes the convergence blows up and even the value
> in the .CHR file changes a lot. In these calculations, while doing the 
> siesta run for
> the scattering region of the system I just put the spin-polarisation flag
> to true, I didnot specify the spin on the parts of the leads in the
> scattering region to be antiparallel by hand. Do I need to do that?
>
>
>
> Please help me regarding this.
>
> thanking you,
>
> Kanchan Ulman
>
>
> -- 
> Kanchan A. Ulman
> Theoretical Sciences Unit
> Jawaharlal Nehru Centre for Advanced Scientific Research
> Jakkur, Bangalore - 560064
> ------------------------------------------------------------------------
>
> _______________________________________________
> Smeagol-discuss mailing list
> Smeagol-discuss at lists.tchpc.tcd.ie
> http://lists.tchpc.tcd.ie/listinfo/smeagol-discuss