ABINIT Discussion Forums

Dear all,

When doing response calculations (phonons, Born effective charge or else) these errors appears really often on a large number of iterations steps:

cgwf3.F90:701:WARNING
New trial energy at line 4 = -4.144536E+02
is higher than former: -4.144536E+02

I can have more than 1000 of them in my log file.

I suppose that this is due to the conjugate gradient algorithms that has trouble finding a minimum. Is there any rules of thumbs to improve the calculation and limit the number of cgwf3 warnings?

Maybe by increasing the variable nline to 5 or 6?

Best regards,

Samuel.

nline might help, but usually it's not a big problem: the energy is practically converged, but in some directions abinit can not get a lower energy by following the gradient calculated for one wf.

See if it finally converges...

Matthieu