> you will need for sure to modify routine sumkt as in the attached > file. Not sure this will solve all problems, though > > P.
Dear Paolo, I managed to solve the problem. There are two routines which were modified, namely, "sumkt" (as you told), and "tweights" (both attached). The problem was that the case nspin=4 (i.e. noncolin=.true.) was treated not properly: there was a loop do ns=1,nspin ... enddo Now I replaced it by (in the spirit of dost.f90): if (nspin==4) then nspin0=1 else nspin0=nspin endif do ns=1,nspin0 ... enddo I tested the code with these modifications, it works correctly. This bug is still present in the latest cvs version of espresso. So, now it can be removed. Thank you very much for your help! Best regards, Iurii Timrov Iurii TIMROV Doctorant (PhD student) Laboratoire des Solides Irradies Ecole Polytechnique F-91128 Palaiseau +33 1 69 33 45 08 timrov at theory.polytechnique.fr -------------- next part -------------- A non-text attachment was scrubbed... Name: sumkt.f90 Type: text/x-fortran Size: 3009 bytes Desc: not available Url : http://www.democritos.it/pipermail/pw_forum/attachments/20110307/dbfe6930/attachment.bin -------------- next part -------------- A non-text attachment was scrubbed... Name: tweights.f90 Type: text/x-fortran Size: 6818 bytes Desc: not available Url : http://www.democritos.it/pipermail/pw_forum/attachments/20110307/dbfe6930/attachment-0001.bin