On Tue, 2011-11-15 at 17:01 -0500, Hanghui Chen wrote: > I find that at the end of relaxation, the code does a new > self-consistent calculation. However, if I turn on Hubbard U, > the code crashes at the last self-consistent calculation due > to some segmentation fault. If no Hubbard U is used, then > everything is fine.
I vaguely remember something like this, but nobody has posted a sample input that exhibits such behavior, can be run in a reasonable time (i.e. few minutes) on a reasonable machine (i.e. a PC). If you have such an example, please post it, or even better, file a bug report on qe-forge: http://qe-forge.org/tracker/?atid=133&group_id=10&func=browse P. -- Paolo Giannozzi, IOM-Democritos and University of Udine, Italy