DFT convergence issues

Just Got Here

12:05:42 AM PDT - Wed, Jul 27th 2016

Dear Community users,
I am trying to perform a qmd calculation in nwchem, however, the job finishes with error. I will be thankful if any one can suggest some help. Also i would like to know whether there are methods such as congugate gradient or steepest descent incorporated in dft module of nwchem. If yes kindly let me know how to invoke them.

My input is as under

start qmd_fe3c
echo
geometry noautosym noautoz
Fe -0.000893 -1.237436 0.078510
Fe 1.206804 0.469766 -0.121583
Fe -1.206122 0.471174 -0.121584

C     0.000917     1.284818     0.713513

end
basis
Fe library lanl2dz_ecp
C library 6-31G*
end

ecp
Fe library lanl2dz_ecp
end

dft

 odft

 xc pbe0

 maxiter 999

convergence energy 1d-3

direct

end

qmd

 nstep_nucl  40000

 dt_nucl     10.0

 targ_temp   600.0

end
task dft qmd

and the error is

This type of error is most commonly associated with calculations not reaching

convergence criteria

 current input line :

    0:

------------------------------------------------------------------------

------------------------------------------------------------------------

This type of error is most commonly associated with calculations not reaching

convergence criteria

------------------------------------------------------------------------

For more information see the NWChem manual at

http://nwchemgit.github.io/index.php/NWChem_Documentation

    0:

------------------------------------------------------------------------

------------------------------------------------------------------------

This type of error is most commonly associated with calculations not reaching

convergence criteria

------------------------------------------------------------------------

For more information see the NWChem manual at

http://nwchemgit.github.io/index.php/NWChem_Documentation

For further details see manual section:

    0:

Forum Regular

7:30:49 AM PDT - Wed, Jul 27th 2016
I was able to converge the SCF for the initial geometry that you provided with your input parameters. Given that, my guess is that during your dynamics you reached a troublesome geometry. This guess is based on you molecule not being optimized and your target temperature being 600 K. If you are say 0.1 Hartree above the minimum on the potential energy surface, this would make your effective temperature over 10,000 K, at those temperatures it is very likely, especially with a small system, that you could reach a geometry that is going to be very difficult, if not practically impossible, to converge the SCF. So I would suggest optimizing your geometry first and then trying the dynamics so as to have better control over the temperature of the system (or use one of the thermostats at least to equilibrate the system first). To use the conjugate gradient algorithm instead of the DIIS algorithm for the SCF, add the keyword cgmin to the DFT block.

Forum >> NWChem's corner >> General Topics