[Wien] Memory problem in Band structure calculation with spin-orbit

Fhokrul Islam Fhokrul.Islam at hik.se
Sun Oct 4 15:10:36 CEST 2009


Hi Prof Blaha and Wien users,

     I previously wrote you about a problem I was having with band structure calculation of semiconducting
supercell with transition metal impurity when spin-orbit coupling is added. As you pointed out the mistake,
I resubmitted the job and, although it did not crash, this time it ran into a different problem. After running
the job for few hours the size of the files in scratch directory started to increase, eventually exceeding
100 Gb !! It doesn't seem reasonable to me since I have calculated the band structure of the same system
without spin-orbit coupling but otherwise with the same set of parameters and that worked out fine. Since 
it didn't produce any error file I am not sure where to look for the problem and how to fixed it. 

My system is a 64 atom GaAs supercell with only one impurity and I am using Wien2k_9_1 version. Only difference 
between the case of without spin-orbit and with spin-orbit is that the number of inequivalent atoms are 10
and 16, respectively. I should mention that the band structure with spin-orbit coupling of a regular GaAs cell
works fine. So it seems like the memory problem arising only with the spin-orbit calculation of the supercell
and I am not sure why it should eat up all the disk space in the scratch directory. Any suggestion will be
appreciated.

Thanks,
Fhokrul     


More information about the Wien mailing list