Thanks for your reply, professor. I found it is dependent on the PPs i chose. For Fe system, if i used the PP: Fe.pbe-nd-rrkjus.UPF, the calculation would stop in this case. But if i used Fe.pbe-sp-van.UPF, it was always going. Then, i think the PPs has significant influence.
Merry Christmas and happy new year. :) Jiayu > hard to say. The problem is in subspace parallel diagonalization (or > in iterative > orthonormalization for Car-Parrinello dynamics). It means that one of > the LAPACK > algorithms used in the parallel algorithm returns an error message. > It may be due > to an ill-conceived algorithm that fails in some cases, to an ill- > conditioned matrix, to > the quality of mathematical libraries...hard to say > ------------------------------ ------------------------------------------- Jiayu Dai Department of Physics National University of Defense Technology, Changsha, 410073, P R China -----------------------------------------