optimization with light atoms


Gets Around
I can not get the optimization with light atoms. I think that this is a numerical effect, but I could not find an option for a more accurate calculation. For example:

echo
start nw
charge 0

geometry   
      He  -8.67636000  -0.30490000   0.00000000   # 1
      He  -5.71316000  -0.37225000   0.00000000   # 2
end 

basis spherical
 * library cc-pvtz
end 

scf 
 maxiter 1000 
end 

dft 
 cgmin 
 xc pbe0
 disp vdw 3 
end 

driver 
 clear 
 TIGHT
 eprec   1e-06
 maxiter 2000
end 

task DFT optimize

Forum Vet
The potential curve for the helium dimer is very flat and our default geometry optimizer is not very smart under these circumstances.
Please try the input below
echo
start nw
charge 0

geometry   
      He  -8.67636000  -0.30490000   0.00000000   # 1
      He  -5.71316000  -0.37225000   0.00000000   # 2
end 

basis spherical
 * library cc-pvtz
end 

scf 
 maxiter 1000 
end 

dft 
 xc pbe0
 disp vdw 3 
grid fine
convergence energy 1d-8
end 

driver 
 clear 
 TIGHT
 eprec   1e-08
 maxiter 2000
 lbfgs
end 


task DFT optimize

Gets Around
I see -very flat.
Unfortunately it does not work. Not even works:
grid xfine
convergence energy 1d-10
end 

driver 
 eprec   1e-10

Gets Around
Maybe the optimizer have some undocumented features to raise the numerical accuracy?

Forum Vet
Could you please provide details about the failure you observe with the input file I have posted (since you report a truncated input file snippet ...)

Gets Around
Sorry, I made a mistake. Your input really works great. Thanks for the help.


Forum >> NWChem's corner >> General Topics