Linear Dependencies


Click here for full thread
Forum Vet
1:41:57 PM PDT - Mon, Apr 9th 2012
That NWChem found linear dependencies is perfectly normal. They are removed from the calculation to provide better convergence.

Being off in the integrated density suggests a not so good starting guess. This should get better once 1) DFT gets closer to convergence, and 2) when DFT optimize gets closer to the real optimal geometry.

Bert



Quote:Roo464 Mar 29th 7:14 pm
Hey All,

I'm new to computational chemistry in general, and NWChem in particular, so I would really appreciate any help/explanation with this error and how to proceed! I've posted the input and excerpts from the output file. I killed the job before it finished because I've been having this problem consistently.

Input:
echo

start molecule

title "imide"
charge 0

geometry units angstroms print xyz autosym noautoz
C           1.599748   -0.446301   -0.001404

C           0.720226    0.632008   -0.007068

C          -0.720251    0.631984   -0.006908

C          -1.599736   -0.446351   -0.000700

C           1.307492   -1.893255    0.005910

C          -1.307434   -1.893291    0.007268

N           0.000037   -2.380608    0.009392

C           2.972337   -0.044555   -0.003498

C           3.137575    1.305552   -0.010945

S           1.616880    2.138796   -0.015252

C          -2.972340   -0.044652   -0.002928

C          -3.137623    1.305446   -0.010823

S          -1.616957    2.138743   -0.014893

H           3.776811   -0.767511    0.000378

H           4.058488    1.871243   -0.014193

H          -3.776789   -0.767634    0.001190

H          -4.058556    1.871105   -0.014231

O           2.217089   -2.714440    0.010880

H           0.000055   -3.396828    0.014434

O          -2.217004   -2.714512    0.011160
end

basis spherical
 * library aug-cc-PVDZ
end

dft
xc xbnl07 0.90 lyp 1.00 hfexch 1.00
cam 0.250000 cam_alpha 0.0 cam_beta 1.0
direct
iterations 500
end

task dft energy



Excerpts from output:

"WARNING : Found 7 linear dependencies
S eigenvalue threshold:     1.00000E-05

Smallest S eigenvalue :     2.26478E-06

Largest  S eigenvalue :     7.40825E-06




!! The overlap matrix has   7 vectors deemed linearly dependent with

   eigenvalues:

2.26D-06 2.90D-06 4.85D-06 5.08D-06 6.87D-06 6.92D-06 7.41D-06"

"WARNING: error on integrated density = 0.24D-03
greater than required accuracy of  0.10D-05"

Thank you!