After doing the convergence test, 'vc-relax' and 'scf' calculations for my material using Quantum ESPRESSO everything went well. But when I wanted to do a 'bands' calculation for bandstructure I got the message Error in routine cdiaghg (283):S matrix not positive definite in the output file. So I increased 'diago_david_ndir' to '4', the default was '2'.And the problem was fixed. But I'm just wondering if I have to repeat the whole calculations from the beginning with diago_david_ndir=4 ( Convergence test, vc-relax, scf) then perform the 'bands' calculations.



You must log in to answer this question.