Simple dimer DFT calculation with diffuse basis set
Clicked A Few Times
8:31:49 AM PDT - Wed, Jul 4th 2018
Hi all!

I've been having a lot of trouble while trying some basis DFT energy calculations on molecular dimers using diffuse basis sets.

This is a quite simple case: the pentacene cofacial stack (dx.doi.org/10.1021/ja061827h)

echo
title "pentacene_b3lyp_6-31+gs"
start  pentacene_b3lyp_6-31+gs

geometry noautoz noautosym nocenter
C	0.000000	0.000000	1.406683
C	0.000000	1.224532	0.727541
C	0.000000	2.463824	1.406276
C	0.000000	3.672444	0.726656
C	0.000000	4.934586	1.408386
C	0.000000	6.108003	0.715689
C	0.000000	0.000000	-1.406683
C	0.000000	1.224532	-0.727541
C	0.000000	2.463824	-1.406276
C	0.000000	3.672444	-0.726656
C	0.000000	4.934586	-1.408386
C	0.000000	6.108003	-0.715689
C	0.000000	-1.224532	-0.727541
C	0.000000	-2.463824	-1.406276
C	0.000000	-3.672444	-0.726656
C	0.000000	-4.934586	-1.408386
C	0.000000	-6.108003	-0.715689
C	0.000000	-1.224532	0.727541
C	0.000000	-2.463824	1.406276
C	0.000000	-3.672444	0.726656
C	0.000000	-4.934586	1.408386
C	0.000000	-6.108003	0.715689
H	0.000000	0.000000	2.492435
H	0.000000	2.464311	2.492092
H	0.000000	4.935578	2.493484
H	0.000000	7.053599	1.245772
H	0.000000	0.000000	-2.492435
H	0.000000	2.464311	-2.492092
H	0.000000	4.935578	-2.493484
H	0.000000	7.053599	-1.245772
H	0.000000	-2.464311	-2.492092
H	0.000000	-4.935578	-2.493484
H	0.000000	-7.053599	-1.245772
H	0.000000	-2.464311	2.492092
H	0.000000	-4.935578	2.493484
H	0.000000	-7.053599	1.245772
C	3.500000	0.000000	1.406683
C	3.500000	1.224532	0.727541
C	3.500000	2.463824	1.406276
C	3.500000	3.672444	0.726656
C	3.500000	4.934586	1.408386
C	3.500000	6.108003	0.715689
C	3.500000	0.000000	-1.406683
C	3.500000	1.224532	-0.727541
C	3.500000	2.463824	-1.406276
C	3.500000	3.672444	-0.726656
C	3.500000	4.934586	-1.408386
C	3.500000	6.108003	-0.715689
C	3.500000	-1.224532	-0.727541
C	3.500000	-2.463824	-1.406276
C	3.500000	-3.672444	-0.726656
C	3.500000	-4.934586	-1.408386
C	3.500000	-6.108003	-0.715689
C	3.500000	-1.224532	0.727541
C	3.500000	-2.463824	1.406276
C	3.500000	-3.672444	0.726656
C	3.500000	-4.934586	1.408386
C	3.500000	-6.108003	0.715689
H	3.500000	0.000000	2.492435
H	3.500000	2.464311	2.492092
H	3.500000	4.935578	2.493484
H	3.500000	7.053599	1.245772
H	3.500000	0.000000	-2.492435
H	3.500000	2.464311	-2.492092
H	3.500000	4.935578	-2.493484
H	3.500000	7.053599	-1.245772
H	3.500000	-2.464311	-2.492092
H	3.500000	-4.935578	-2.493484
H	3.500000	-7.053599	-1.245772
H	3.500000	-2.464311	2.492092
H	3.500000	-4.935578	2.493484
H	3.500000	-7.053599	1.245772
end

basis
 * library "6-31+g*"
end

dft
 direct
 convergence diis 3
 maxiter 1000
 grid fine
 xc b3lyp
 vectors input atomic output dimer.movecs
end
task dft energy


The SCF cycle starts like that:
d= 0,ls=0.0,diis     1  -1638.2924516151 -6.63D+03  2.27D+02  4.95D+03   916.0
 Grid integrated density:     291.997207182703
 Requested integration accuracy:   0.10E-06
 d= 0,ls=0.5,diis     2  -1596.5938570357  4.17D+01  9.24D+01  4.04D+02  1147.7
 Grid integrated density:     292.000933360791
 Requested integration accuracy:   0.10E-06
 d= 0,ls=0.5,diis     3   -533.7879528180  1.06D+03  7.82D+00  5.05D+03  1377.1
 Grid integrated density:     292.002854672656
 Requested integration accuracy:   0.10E-06
 d= 0,ls=0.5,diis     4  -1198.9782229509 -6.65D+02  1.97D+02  2.15D+03  1606.3
 Grid integrated density:     292.001966967721
...


and eventually just fails to converge:

 d= 0,ls=0.5,diis    29   1352.8835222278 -1.60D+03  2.90D+00  7.18D+03  6620.0
 Grid integrated density:     292.000050022126
 Requested integration accuracy:   0.10E-06
 d= 0,ls=0.5,diis    30   1254.6589767786 -9.82D+01  4.28D+00  5.78D+03  6825.0
 Grid integrated density:     292.002613695296
 Requested integration accuracy:   0.10E-06
 d= 0,ls=0.5,diis    31   3773.8405074356  2.52D+03  5.69D+00  5.06D+03  7029.3


I've also tried the fragment MO method and successive projections from smaller basis (3-21g to 6-31g to 6-31g* to 6-31+g*) but nothing works...

Any input will be highly appreciated!

Best regards,
Orestis

Forum Vet
9:42:28 AM PDT - Fri, Jul 6th 2018
Please try this modified input that will converge in 18 iterations

echo
title "pentacene_b3lyp_6-31+gs"
start  pentacene_b3lyp_6-31+gs

geometry noautoz 
C	0.000000	0.000000	1.406683
C	0.000000	1.224532	0.727541
C	0.000000	2.463824	1.406276
C	0.000000	3.672444	0.726656
C	0.000000	4.934586	1.408386
C	0.000000	6.108003	0.715689
C	0.000000	0.000000	-1.406683
C	0.000000	1.224532	-0.727541
C	0.000000	2.463824	-1.406276
C	0.000000	3.672444	-0.726656
C	0.000000	4.934586	-1.408386
C	0.000000	6.108003	-0.715689
C	0.000000	-1.224532	-0.727541
C	0.000000	-2.463824	-1.406276
C	0.000000	-3.672444	-0.726656
C	0.000000	-4.934586	-1.408386
C	0.000000	-6.108003	-0.715689
C	0.000000	-1.224532	0.727541
C	0.000000	-2.463824	1.406276
C	0.000000	-3.672444	0.726656
C	0.000000	-4.934586	1.408386
C	0.000000	-6.108003	0.715689
H	0.000000	0.000000	2.492435
H	0.000000	2.464311	2.492092
H	0.000000	4.935578	2.493484
H	0.000000	7.053599	1.245772
H	0.000000	0.000000	-2.492435
H	0.000000	2.464311	-2.492092
H	0.000000	4.935578	-2.493484
H	0.000000	7.053599	-1.245772
H	0.000000	-2.464311	-2.492092
H	0.000000	-4.935578	-2.493484
H	0.000000	-7.053599	-1.245772
H	0.000000	-2.464311	2.492092
H	0.000000	-4.935578	2.493484
H	0.000000	-7.053599	1.245772
C	3.500000	0.000000	1.406683
C	3.500000	1.224532	0.727541
C	3.500000	2.463824	1.406276
C	3.500000	3.672444	0.726656
C	3.500000	4.934586	1.408386
C	3.500000	6.108003	0.715689
C	3.500000	0.000000	-1.406683
C	3.500000	1.224532	-0.727541
C	3.500000	2.463824	-1.406276
C	3.500000	3.672444	-0.726656
C	3.500000	4.934586	-1.408386
C	3.500000	6.108003	-0.715689
C	3.500000	-1.224532	-0.727541
C	3.500000	-2.463824	-1.406276
C	3.500000	-3.672444	-0.726656
C	3.500000	-4.934586	-1.408386
C	3.500000	-6.108003	-0.715689
C	3.500000	-1.224532	0.727541
C	3.500000	-2.463824	1.406276
C	3.500000	-3.672444	0.726656
C	3.500000	-4.934586	1.408386
C	3.500000	-6.108003	0.715689
H	3.500000	0.000000	2.492435
H	3.500000	2.464311	2.492092
H	3.500000	4.935578	2.493484
H	3.500000	7.053599	1.245772
H	3.500000	0.000000	-2.492435
H	3.500000	2.464311	-2.492092
H	3.500000	4.935578	-2.493484
H	3.500000	7.053599	-1.245772
H	3.500000	-2.464311	-2.492092
H	3.500000	-4.935578	-2.493484
H	3.500000	-7.053599	-1.245772
H	3.500000	-2.464311	2.492092
H	3.500000	-4.935578	2.493484
H	3.500000	-7.053599	1.245772
end

basis spherical
 * library "6-31+g*"
end

dft
 direct
# convergence diis 3
 maxiter 1000
 grid fine
 xc b3lyp
 vectors input atomic output dimer.movecs
 smear
 convergence ncydp 0 damp 45 dampon 1d99 dampoff 1d-4
end
set quickguess t
task dft energy

Clicked A Few Times
8:06:00 AM PDT - Sat, Jul 7th 2018
Dear Edo,

Thanks a lot for the input! I found the options:
smear
convergence ncydp 0 damp 45 dampon 1d99 dampoff 1d-4

in another post in the forum and tried them successfully!

Moreover, while searching for solutions, I managed to get convergence using another approach: I just ran a RHF SCF single point energy calculation and resume with the obtained movecs:
scf
 direct
 maxiter 500
 vectors input atomic output hfd.movecs
end
task scf energy
dft
 direct
 convergence diis 4
 maxiter 500
 grid fine
 xc b3lyp
 vectors input hfd.movecs output dimer.movecs
end
task dft energy


By the way, it would be really nice if NWCHEM documentation includes an entry for the ET FMO method. The only guidance I had on this very helpful module was from the etrans-fmo QA example...

Again, many thanks!

Best regards,
Orestis


Forum >> NWChem's corner >> General Topics