TURBOMOLE Users Forum
TURBOMOLE Modules => Ricc2 => Topic started by: mak_108 on January 29, 2020, 03:37:36 PM
-
Hello,
I'm trying to perform ricc2 (sos-cc2) excited state calculations (energy calculation for 5 states and geometry optimization of the lowest one) for a system of 58 atoms. Unfortunately, I get the following error:
==========================================
======== CC DENSITY MODULE ========
current wave-function model: CC2
calculating 3 eta densities
the non-canonical algorithm will be used for all densities
density nr. cpu/min wall/min L R
------------------------------------------------------
========================
internal module stack:
------------------------
ricc2
cc_rspdrv
cc_rspden
cc_1den
ccintrightvec
========================
fatal error for MPI rank 1
Missing Laplace input for cc_1den with R1 or RE1.
ricc2 ended abnormally
error in gradient step (1)
========================================
What does Laplace input for cc_1den refer to??
The same routine worked quite well for a system of 20 atoms, so my immediate suspicion was that it might have something to do with memory.
Any help would be greatly appreciated.
Thanks,
Mallikarjun
-
Dear Mallikarjun,
can you also provide your control file? Judging from the error message the $laplace data group is missing, which is needed for SOS-CC2 and SOS-MP2. Add for example
$laplace
conv=5
to your control file and see if your calculation then goes through. If still the same error message pops up, it is likely a bug and I suggest to contact the Turbomole support. I also suggest that your read chapter "9.6 Laplace-transformed SOS-RI-MP2..." in the Turbomole documentation for further information. It briefly describes how the "Laplace-Trick" is used to replace the energy denominator which an equivalent Laplace transformed expression
-
Thank you for the kind reply. I realised that, and am now running two jobs, one with conv=4 and the other with conv=5. If this does not work, would it make sense to completely get rid of the SOS and just try to see if CC2 without SOS works?
Here is my new control file:
$title
$operating system unix
$symmetry c1
$redundant file=coord
$user-defined bonds file=coord
$coord file=coord
$optimize
internal on
redundant on
cartesian off
global off
basis off
$atoms
c 1-4,7,9-11,13,15-17,19-21,24-40 \
basis =c aug-cc-pVDZ \
cbas =c aug-cc-pVDZ
n 5-6,8,12,14,18,22-23 \
basis =n aug-cc-pVDZ \
cbas =n aug-cc-pVDZ
h 41-58 \
basis =h aug-cc-pVDZ \
cbas =h aug-cc-pVDZ
$basis file=basis
$rundimensions
dim(fock,dens)=677371
natoms=58
nshell=450
nbf(CAO)=1162
dim(trafo[SAO<-->AO/CAO])=1322
rhfshells=1
nbf(AO)=1082
$scfmo file=mos
$closed shells
a 1-133 ( 2 )
$scfiterlimit 130
$scfconv 7
$thize 0.10000000E-04
$thime 5
$scfdamp start=0.300 step=0.050 min=0.100
$scfdump
$scfintunit
unit=30 size=0 file=twoint
$scfdiis
$maxcor 5000 MiB per_core
$scforbitalshift automatic=.1
$drvopt
cartesian on
basis off
global off
hessian on
dipole on
nuclear polarizability
$interconversion off
qconv=1.d-7
maxiter=25
$coordinateupdate
dqmax=0.3
interpolate on
statistics 5
$forceupdate
ahlrichs numgeo=0 mingeo=3 maxgeo=4 modus=<g|dq> dynamic fail=0.3
threig=0.005 reseig=0.005 thrbig=3.0 scale=1.00 damping=0.0
$forceinit on
diag=default
$energy file=energy
$grad file=gradient
$forceapprox file=forceapprox
$denconv 0.10000000E-06
$cbas file=auxbasis
$ricc2
hard_restart
cc2
geoopt model=cc2 state=(s1)
sos cos= 1.30000
$response
fop unrelaxed_only operators=diplen
sop operators=(diplen,diplen)
$excitations
irrep=a nexc=5
$laplace
conv=5
$last step ricc2
$last SCF energy change = 0.87090939E-08
$charge from dscf
0.000 (not to be modified here)
$dipole from dscf
x -0.00034122219716 y 0.00013060160870 z -0.00003075498512 a.u.
| dipole | = 0.0009319488 debye
$orbital_max_rnorm 0.12832582988654E-03
$excitation_energies_CCS_____1^a___ file=exstates
$excitation_energies_CC2_____1^a___ file=exstates
$t2_norm_of_re0_CC2_____1^a___ file=exstates
$t2_norm_of_le0_CC2_____1^a___ file=exstates
$<le0|re0>-overlap_CC2_____1^a___ file=exstates
$<le01|re01>-overlap_CC2_____1^a___ file=exstates
$<le02|re02>-overlap_CC2_____1^a___ file=exstates
$end
-
If you get the SOS-CC2 calculation through in small time, you can of course also try to do CC2. SOS-CC2 scales as O(N^4), while standard CC2 scales as O(N^5). So there are computational benefits for larger systems. For smaller systems I would suggest using CC2 and for small systems SOS-CC2 might even be slower due to a larger prefactor.
In the manual you can find for SOS-MP2 an estimator for the break even point of SOS-MP2 and MP2. If the number of your occupied orbitals are 6 times larger than the number of your Laplace points you can start to expect savings. Maybe this also helps you in your considerations.
-
Thank you. I'll leave an update once something works, for future reference of other users.