DFT: Convergence issue CH4


Just Got Here
Hi,

I'm a new user of NWChem. Therefore I'm trying to run some dummy systems to get my hand into it.
I'm trying with a CH4 molecule. For example, the following inputs work fine and nwchem finish successfully.

geometry noautoz noautosym
  C     7.67039    10.29926     4.07353
  H     7.63303    10.29931     5.15228
  H     8.00346    11.27118     3.73943
  H     8.30423     9.51459     3.64992
  H     6.65039    10.11495     3.79618
end

charge 0
scratch_dir /home/marc/Documents/simbeyond/votca/tuto-gromcas/tuto7/rundir028/tmp/nwchem/id_1
start
basis
 * library 3-21g   
end
memory 1500 mb

dft
 mult 1
 print "ao overlap"
 xc xpbe96 cpbe96
 direct
 noprint "final vectors analysis" 
 maxiter 100
end
task dft



However, the following code, which is identical except that the molecule is translated to a new position, fails to converge

geometry noautoz noautosym
  C     7.02375    12.76341    11.24170
  H     7.58887    13.56485    10.79067
  H     6.45492    13.16440    12.06805
  H     6.36312    12.23783    10.54583
  H     7.78718    12.08897    11.57874
end

charge 0
scratch_dir /home/marc/Documents/simbeyond/votca/tuto-gromcas/tuto7/rundir028/tmp/nwchem/id_4
start
basis
 * library 3-21g   
end
memory 1500 mb

dft
 mult 1
 print "ao overlap"
 xc xpbe96 cpbe96
 direct
 noprint "final vectors analysis" 
 maxiter 100
end
task dft



I tried to increase the number of iterations and played a bit around with convergence parameters, but I do not manage to get the scf loop to converge. How is it possible that this system which is almost identical to the previous one do not converge? What do I'm missing?

Thanks in advance for your help,
Marc Barbry

Forum Vet
Marc
I am not able to reproduce your failure withe NWChem 6.8
Could you upload your output files to a public website?

Forum Vet
Sure, both inputs produced results.

You can try it by yourself, because I have already discarded the log files.

Just Got Here
Hi,

Thanks for your answers.
You can download the output files with the following link.
[1]

I'm using nwchem 6.6, but I guess that it should not matter so much for such basic calculations.

Marc

Forum Vet
Marc
Both of your output files are numerically incorrect. This seems to point to issues in your installation.
Could you try to install NWChem 6.8 by following the instructions at
https://github.com/nwchemgit/nwchem/wiki/Compiling-NWChem


Forum >> NWChem's corner >> Running NWChem