On Fri, September 18, 2009 06:37, Haowei Peng wrote: > However, the relax calculation using similar parameters and input file > goes well.
Dear Haowei Peng, your input looks ok, apart that you may have missed to set lda_plus_U to true (or you just wanted to try the calculation without the U) > lda_plus_u=.false. Hubbard_U(1)=1.d-8 > diagonalization (ZHEGV*) failed I'm having this error a lot on a specific hardware with a specific version of ifort+acml libraries. On the same hardware, with mkl I can't even finish the first iteration because of those diagonalization errors. I know it's not a great advice, but I would try a different version of the compiler, just to be sure. best regards -- Lorenzo Paulatto SISSA & DEMOCRITOS (Trieste) phone: +39 040 3787 511 skype: paulatz www: http://people.sissa.it/~paulatto/ ---------------------------------------------------------------- SISSA Webmail https://webmail.sissa.it/ Powered by SquirrelMail http://www.squirrelmail.org/