iOpenShell » Q-Chem specific questions » SF-TDDFT analytical gradients

SF-TDDFT analytical gradients

Page: 1

Author Post
Member
Registered: Apr 2016
Posts: 3
I am calculating analytical gradients using SF-TDDFT and noticed there is a large difference between the analytical gradients and the numerical gradients.

For example, for ethene the numerical gradient is:
0.0018073 -0.0000001 0
-0.0018074 0.0000001 0
0.0009267 0.0019783 0
0.0009267 -0.0019783 0
-0.0009267 0.0019783 0
-0.0009267 -0.0019783 0

However, the analytical gradient is:
-0.0045603 0 0.0000002
0.0045603 0 -0.0000001
0.0008078 0.000518 0
0.0008078 -0.000518 -0.0000001
-0.0008078 0.000518 -0.0000001
-0.0008078 -0.000518 0.0000001

I've tried specifying that the CIS_RELAXED_DENSITY TRUE however, that didn't change the values much.

Any tips? Is SF-TDDFT gradients implemented?
Member
Registered: Apr 2016
Posts: 3
Hello all, I'm still having trouble with this problem. I've attached my input file to help trouble shoot. Note that the results I posted before used a larger basis set cc-PVDZ and the numerical and analytical gradient are sensitive to basis set.

$rem
JOBTYPE FORCE
UNRESTRICTED TRUE
EXCHANGE PBE0
BASIS 6-31G*
SPIN_FLIP 1
RPA FALSE
SCF_CONVERGENCE 8
CIS_N_ROOTS 4
CIS_CONVERGENCE 10
MAX_CIS_CYCLES 100
CIS_STATE_DERIV 1
symmetry off
SYM_IGNORE TRUE
$end

$molecule
0 3
C -0.550647 1.045151 0.000060
C 0.776937 1.045151 0.000060
H -1.118883 1.087285 -0.924706
H -1.118883 1.003016 0.924826
H 1.345173 1.087285 -0.924706
H 1.345173 1.003016 0.924826
$end


Also, the analytical gradient depends on the number of CIS_N_ROOTS (which is very strange):
e.g.
4 roots
0.0040703 -0.0040703 0.0012195 0.0012195 -0.0012195 -0.0012195
0.0000001 0.0000001 -0.0000205 0.0000205 -0.0000205 0.0000205
0 0 0.0002697 -0.0002697 0.0002697 -0.0002697
8 roots
0.0023568 -0.0023568 0.0011217 0.0011217 -0.0011217 -0.0011217
0.0000001 0.0000001 -0.0000113 0.0000113 -0.0000113 0.0000113
0 0 0.0000711 -0.0000711 0.0000711 -0.0000711
12 roots
0.0042112 -0.0042112 0.000953 0.000953 -0.000953 -0.000953
0.0000001 0.0000001 0.0000052 -0.0000053 0.0000052 -0.0000053
0 0 -0.0002973 0.0002973 -0.0002973 0.0002973
Member
Registered: Apr 2016
Posts: 3
The problem was a bug in the code. Yihan fixed it. Thanks Yihan.

-C

Page: 1

iOpenShell » Q-Chem specific questions » SF-TDDFT analytical gradients