CFOUR
Use serial CFOUR 2.1
input
Input file need to be called ZMAT
Documentation : http://slater.chemie.uni-mainz.de/cfour/index.php?n=Main.InputFileZMAT
ZMAT input example for acetone
Acetone CC3/aug-cc-pVTZ S0 optimised geometry CC2/VDZ calculation
C 0.00000000 0.00000000 0.18807702
C 0.00000000 2.42007545 -1.31764698
C 0.00000000 -2.42007545 -1.31764698
O 0.00000000 0.00000000 2.48269094
H 0.00000000 4.03690733 -0.05185132
H 0.00000000 -4.03690733 -0.05185132
H 1.66061256 2.48420530 -2.53995285
H -1.66061256 2.48420530 -2.53995285
H 1.66061256 -2.48420530 -2.53995285
H -1.66061256 -2.48420530 -2.53995285
*CFOUR(COORD=CARTESIAN,CALC=CC2,BASIS=PVDZ,EXCITE=EOMEE
ESTATE_SYM=1/0/2/2,ESTATE_CONV=8,SCF_CONV=8,CC_CONV=8
FROZEN_CORE=ON)
For big memory calculation, add and ajust last line containing *CFOUR :
*CFOUR(…, MEMORY=16,MEM_UNIT=GB)
sbatch
Thereis a slurm sbatch example.
We ask here for 20GB of RAM, CFOUR consume 762MB by himself. For small calculations --mem can be reduced.
#!/bin/bash
#SBATCH -J cfour
#SBATCH -N 1
#SBATCH -n 1
#SBATCH --ntasks-per-core=1
#SBATCH --ntasks-per-node=1
#SBATCH --mem=20000
#SBATCH -p xeonv4_mono
#SBATCH --gres=ioperso
# Load CFOUR v2.1 module
module load cfour/v2.1-64bits-serial
# output file name
file="acetone_cc2_vdz_S0min_exc"
# copy input file in TMPDIR
cp ZMAT $TMPDIR
# run calculation
cd $TMPDIR
time xcfour > ${file}.out
# copy output in submit directory
cp ${file}.out $SLURM_SUBMIT_DIR
Use parallel CFOUR 2.1
Same as serial version, just change module load :
module load cfour/v2.1-64bits-parallel
As well as cpu/mémoire/etc
DO NOT USE mpirun/mpiexec/autre, xcfour detect variables