[QE-users] S matrix not positive definite

[Lorenzo Paulatto]

Hello,
have you put many empty bands in the nscf calculation?
You can only include a limited number of empty bands before running into 
troubles: either because the pseudopotentials are not transferable to 
such high energies, or because you don't have enough plane-waves to 
represent such high energy states.

Otherwise, it may be that your system has a very tiny gap, or no gap at 
all at a specific k-point that was not included in the SCF calculation, 
but is in the NSCF one.

kind regards

On 08/10/2018 10:25, Ubaid Mohd wrote:
> Dear Experts,
>   I am doing a calculation related to bilayer phosphorene. I have 
> successfully done vc relaxation and bandstructure calculations. When I 
> am doing the nscf calculation using the same outdir & prefix flag, I am 
> getting the error.........
> 
>       Computing kpt #:    45
>       total cpu time spent up to now is     4384.4 secs
> 
>       Computing kpt #:    46
>       c_bands:  2 eigenvalues not converged
>       total cpu time spent up to now is     4497.2 secs
> 
>       Computing kpt #:    47
>       c_bands:  2 eigenvalues not converged
>       total cpu time spent up to now is     4611.2 secs
> 
>       Computing kpt #:    48
>       c_bands:  4 eigenvalues not converged
>       total cpu time spent up to now is     4727.0 secs
> 
>       Computing kpt #:    49
>       c_bands:  2 eigenvalues not converged
>       total cpu time spent up to now is     4840.0 secs
> 
>       Computing kpt #:    50
>       c_bands:  2 eigenvalues not converged
>       total cpu time spent up to now is     4946.0 secs
> 
>       Computing kpt #:    51
>       c_bands:  2 eigenvalues not converged
>       total cpu time spent up to now is     5053.2 secs
> 
>       Computing kpt #:    52
> 
>       Error in routine cdiaghg (904):
>       S matrix not positive definite
>        stopping ...
> 
> I search the solution of related problem on google in which error is due 
> to "atoms are very close to each other" but, if atoms are very close to 
> each other, why I am getting the optimized structure and the correct 
> band gap?
> How to sort out this problem to proceed the further calcultions like 
> dos, pdos and charges etc.
> 
> Thanks and regards
> 
> /Mohammad Ubaid/
> /PhD Research Scholar/
> /Department of Physics/
> /Jamia Millia Islamia University/
> /New Delhi - 110025/
> 
> _______________________________________________
> users mailing list
> users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
> 

-- 
Lorenzo Paulatto - Paris