[molpro-user] mcscf problems
Valentina Cantatore
valecantatore at gmail.com
Tue Dec 4 10:24:49 GMT 2012
Dear molpro users,
I am a new trial user of Molpro. I was performing an mcscf optimization for a deformed structure of the azobenzene molecule with the 2010.1 version of the code. The input that I used is:
***TAB (C2)
memory,200, m
gprint,basis,orbitals,civector
file,2,rot1814.sp.optg.wfn
punch,rot1814.optg.pun,new
NXX3 = 0.61995528D+00 ANG
CN5 = 0.14279844D+01 ANG
CNXX5 = 0.11505652D+03 DEGREE
DIH5 = 45.0
CC6 = 0.13908971D+01 ANG
CCN6 = 0.12443263D+03 DEGREE
DIH6 = -0.14065092D+00 DEGREE
CC7 = 0.13844965D+01 ANG
CCN7 = 0.11542246D+03 DEGREE
DIH7 = -0.18013427D+03 DEGREE
HC8 = 0.10742867D+01 ANG
HCC8 = 0.11878147D+03 DEGREE
DIH8 = 0.18000000D+03 DEGREE
HC9 = 0.10720031D+01 ANG
HCC9 = 0.11970712D+03 DEGREE
DIH9 = 0.18000000D+03 DEGREE
CC10 = 0.13817785D+01 ANG
CCC10 = 0.11946866D+03 DEGREE
DIH10 = 0.26434144D-14 DEGREE
CC11 = 0.13865300D+01 ANG
CCC11 = 0.12018910D+03 DEGREE
DIH11 = -0.39103991D-14 DEGREE
HC12 = 0.10749358D+01 ANG
HCC12 = 0.11992410D+03 DEGREE
DIH12 = 0.18000000D+03 DEGREE
HC13 = 0.10751730D+01 ANG
HCC13 = 0.11963628D+03 DEGREE
DIH13 = -0.18000000D+03 DEGREE
CC14 = 0.13894673D+01 ANG
CCC14 = 0.12046391D+03 DEGREE
DIH14 = 0.00000000D+00 DEGREE
HC15 = 0.10752112D+01 ANG
HCC15 = 0.11996380D+03 DEGREE
DIH15 = -0.18000000D+03 DEGREE
Geometry={
ang
q
q, 1, 1.00
n, 1, nxx3, 2, 90.000
n, 1, nxx3, 3, 180.000, 2, 90.000
c, 3, cn5 , 1, cnxx5 , 2, dih5
c, 5, cc6 , 3, ccn6 , 1, dih6
c, 5, cc7 , 3, ccn7 , 1, dih7
h, 7, hc8 , 5, hcc8 , 6, dih8
h, 6, hc9 , 5, hcc9 , 7, dih9
c, 6, cc10, 5, ccc10 , 7, dih10
c, 7, cc11, 5, ccc11 , 6, dih11
h, 11, hc12, 7, hcc12 , 5, dih12
h, 10, hc13, 6, hcc13 , 5, dih13
c, 10, cc14, 6, ccc14 , 5, dih14
h, 14, hc15, 10, hcc15 , 6, dih15
c, 4, cn5 , 1, cnxx5 , 2, dih5
c, 16, cc6 , 4, ccn6 , 1, dih6
c, 16, cc7 , 4, ccn7 , 1, dih7
h, 18, hc8 , 16, hcc8 , 17, dih8
h, 17, hc9 , 16, hcc9 , 18, dih9
c, 17, cc10, 16, ccc10 , 18, dih10
c, 18, cc11, 16, ccc11 , 17, dih11
h, 22, hc12, 18, hcc12 , 16, dih12
h, 21, hc13, 17, hcc13 , 16, dih13
c, 21, cc14, 17, ccc14 , 16, dih14
h, 25, hc15, 21, hcc15 , 17, dih15
}
basis=6-31G*
{rhf;wf,96,1,0}
put,molden,rhf.s0.rot1814.optg.molden
{multi
noextra;
maxiter,40;
closed,20,19;occ,27,26
wf,96,1,0;state,1}
put,molden,rot1814.optg.mcscf.molden
{optg
inactive dih5;}
put,molden,rot1814.optg.molden
rs2c
---
I run this optimization on the cineca supercoputer plx using the following script:
#!/bin/bash
#PBS -l select=1:ncpus=12:mpiprocs=12:mem=47GB
#PBS -l walltime=24:00:00
#PBS -q longpar
#PBS -A *****
export TMPDIR=$CINECA_SCRATCH
cd $PBS_O_WORKDIR
module load autoload molpro
molpro -n12 rot1814.sp.optg.inp
The optimization finished and gave me an energy of ENERGY=-569.15192930. The program wasn't able to run CASPT2 calculation so I took the optimized structure and I performed a single point calculation on another machine using 4 processors and the following input:
***,(C2)
memory,550, m
gprint,basis,orbitals,civector
file,2,rot1814.optg.sp.wfn,new
punch,rot1814.optg.sp.pun,new
geomtyp=xyz
Geometry={
24
MULTI000/6-31G* ENERGY=-569.15192930
N -0.5325246258 -0.4032695280 -1.4214415833
N 0.5325246258 0.4032695280 -1.4214415833
C -1.5570952519 -0.1176059733 -0.5751681918
C -1.5875016866 0.9782914822 0.3240902310
C -2.6828117999 -0.9753265702 -0.6372874957
H -2.6542458744 -1.8035492359 -1.3200190152
H -0.7475337514 1.6417051215 0.3892917572
C -2.7055206902 1.1877869496 1.1168675931
C -3.7858422447 -0.7425057332 0.1659015099
H -4.6345043297 -1.3999893704 0.1103274814
H -2.7223557940 2.0205059709 1.7966704857
C -3.7995693614 0.3358445641 1.0417221045
H -4.6589748201 0.5118515444 1.6631715504
C 1.5570952519 0.1176059733 -0.5751681918
C 1.5875016866 -0.9782914822 0.3240902310
C 2.6828117999 0.9753265702 -0.6372874957
H 2.6542458744 1.8035492359 -1.3200190152
H 0.7475337514 -1.6417051215 0.3892917572
C 2.7055206902 -1.1877869496 1.1168675931
C 3.7858422447 0.7425057332 0.1659015099
H 4.6345043297 1.3999893704 0.1103274814
H 2.7223557940 -2.0205059709 1.7966704857
C 3.7995693614 -0.3358445641 1.0417221045
H 4.6589748201 -0.5118515444 1.6631715504
}
basis=6-31G*
{rhf;wf,96,1,0}
put,molden,rhf.rot1814.optg.sp.molden
{multi
noextra;
maxiter,40;
closed,20,19;occ,27,26
wf,96,1,0;state,1}
put,molden,rot1814.optg.sp.mcscf.molden
rs2c
---
I expected the same convergence value at the end of the mcscf cycles as I got in other previous calculation. But this time was different. After the mcscf cycle I obtained an energy value of -569.15359016. Why I got this difference? The only parameters that changed was the number of the processors and the memory required. Do you have any hint to avoid these instabilities?
I also tried to redo a new optimization starting from the geometry obtained from the first and I god a third new value. What is the value that I need to trust? Can you give me any hint to avoid these instabilities?
Thank you very much for your attention.
Valentina Cantatore
PhD student
Università di Pisa
More information about the Molpro-user
mailing list