GAPW and GTH
Harald Forbert
harald.... at theochem.ruhr-uni-bochum.de
Mon Jun 18 17:27:47 UTC 2007
Teo asked me to also post this here:
When calculating HCl with GTH pseudos, I get weird energies with
GAPW (positive total energy: +11.6) whereas with GPW
it is reasonably bound at -15.5. Now if I turn on FULL_GAPW
it again gives sort of the same energy as GPW, but in both GAPW
cases the conserved quantity is not conserved - even in a few (5)
steps MD I get a change of 1e-3 whereas the corresponding
GPW calculation the conserved quantity only changes by 1.0e-7.
I did not do a longer MD with FULL_GAPW, but a similar
calculation with "normal" GAPW also blew up, so it's not just
a miscalculated conserved quantity.
Both GPW+GTH and GAPW with all electrons seem to work fine.
I used the GTH_BASIS_SETS and GTH_POTENTIALS from
the cvs tree and I pasted the input file below.
Am I doing something wrong, or is GAPW not fully working?
The same sort of setup works perfectly fine for a single water
molecule.....
&GLOBAL
PROGRAM_NAME CP2K
PROJECT gapw
RUN_TYPE MD
PRINT_LEVEL LOW
&END GLOBAL
&MOTION
&MD
ENSEMBLE NVE
STEPS 5
TIMESTEP 0.2
TEMPERATURE 50.0
&END MD
&END MOTION
&FORCE_EVAL
METHOD QS
&DFT
BASIS_SET_FILE_NAME GTH_BASIS_SETS
POTENTIAL_FILE_NAME GTH_POTENTIALS
&QS
EPS_DEFAULT 1.0E-14
EXTRAPOLATION PS
EXTRAPOLATION_ORDER 3
METHOD GAPW
# FULL_GAPW T
MAP_CONSISTENT T
&END QS
&SCF
MAX_SCF 100
EPS_SCF 1.0E-7
CHOLESKY F
SCF_GUESS ATOMIC
&END SCF
&MGRID
CUTOFF 400
&END MGRID
&XC
&XC_FUNCTIONAL BLYP
&END XC_FUNCTIONAL
&END XC
&END DFT
&SUBSYS
&CELL
ABC 8.0 8.0 8.0
UNIT ANGSTROM
&END CELL
&COORD
Cl 0.9564675869 -0.4777215658 0.1818080168
H -0.1657168075 0.1752674377 -0.0159958272
&END COORD
&VELOCITY
1.57689068E-05 2.13541290E-05 1.89890498E-06
-2.92031628E-04 6.55172505E-05 1.82242571E-04
&END VELOCITY
&KIND Cl
BASIS_SET TZV2P-GTH
MASS 3.4968852720999998E+01
POTENTIAL GTH-BLYP-q7
&END KIND
&KIND H
BASIS_SET TZV2P-GTH
MASS 1.0079469999999999E+00
POTENTIAL GTH-BLYP-q1
&END KIND
&END SUBSYS
&END FORCE_EVAL
More information about the CP2K-user
mailing list