On Mer, Luglio 2, 2008 08:56, vega lew wrote: > After several iterations of self-consistent calculation, there is an > error encountered. > %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% > from pzpotrf : error # 1 > problems computing cholesky decomposition > %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
The subroutine is called during parallel diagonalization, so it may be a problem in you mpi setup as much as a problem of QE. If the former is true, you can try to restart the calculation and see what happens. If it fails, you can also try to restart from scratch and see if it stops at the same point. If it stops at the same point, it is probably a bug in the parallel implementation, you can try setting diagonalization='david-serial', which should work around it but is slower for large systems. Please let us know the result of your tests. bye -- Lorenzo Paulatto SISSA & DEMOCRITOS (Trieste) +39 040 3787 511 http://people.sissa.it/~paulatto/ ---------------------------------------------------------------- SISSA Webmail https://webmail.sissa.it/ Powered by SquirrelMail http://www.squirrelmail.org/
