Dear Roberto Thanks for your help. Now the problem has been got settled with your suggestion.
Best regards, Hiro R.C.Pasianot さんは書きました: > > Hi > > Try disabling the "divide and conquer" method of matrix diagonalization > ( Diag.DivideAndConquer .false. ). > Also the Diag.Memory switch might me helpful. > > For new calculations it is often preferable not to use old density > matrices, namely, let SIESTA build a fresh one. > > Bye, > > Roberto > > > On Fri, 8 May 2009, Hiromasa OHNISHI wrote: > >> Dear siesta users >> >> In the calculation with Diag.ParralelOverK option, >> I have got the following error. >> >> Failure to converge standard eigenproblem >> Stopping Program from Node: 7 >> >> This error often occurs in the middle of the scf iteration. >> How to eliminate this error ? >> >> Hiro >> > >