[CPMD-list] Energy conservation etc.
padma at uiuc.edu
padma at uiuc.edu
Tue Nov 1 19:53:59 CET 2005
Dear cpmd users,
I am facing poor energy conservation in CPMD runs of a
CO2+ water system (one CO2 molecule in 46 H2O) with PBC.
(The input file used for the MD is given below.)
The starting configuration was force-field-equilibrated
structure of the system which is then wavefunction optimized
to 10.0d-6.
The conservation (slope) of EHAM over a 5 ps run was
about 10d-4/ps (which, I believe, is atleast one order
poorer than acceptable value). The kinetic energy of electrons
and ions looked pretty good with fluctuations about their
set-values.
Further, hoping that some adjustments in the input parameters
would improve the situation, a shoter run of about 0.2 ps
is made (starting from the RESTART file at the end of
the 5ps run) with the following options:
RESTART LATEST WAVEFUNCTION COORDINATES VELOCITIES NOSEP
QUENCH BO
NOSE IONS
300.0 3200.0
nose electrons
0.02 10000.0
(no MASSIVE thermostating for ions and no thermostating
at all for the electrons)
also in &DFT section the GC-CUTOFF increased
GC-CUTOFF
1.0e-5
But KSH was found greater then EHAM during the whole run!!
And the NOSE ENERGY IONS listed in the AVERAGED QUATITIES at
the end of the run was negative (~ -0.9 ).
I would be very greatful to know what's going wrong
and how can I improve the conservation.
best regards,
Padma
&CPMD
MOLECULAR DYNAMICS CP
optimizE WAVEFUNCTION
RESTART LATEST WAVEFUNCTION COORDINATES VELOCITIES NOSEP NOSEE
CONVERGENCE
1.0e-6 1.0e-6
SPLINE POINTS
1001
TEMPERATURE
300.0
NOSE IONS MASSIVE
300.0 2500.0
NOSE ELECTRONS
0.02 10000.0
MAXSTEP
4000
TRAJECTORY SAMPLE
20
TIMESTEP
4.0
EMASS
600.0
&END
&DFT
NEWCODE
EXCHANGE CORRELATION TABLE NO
FUNCTIONAL BLYP
GC-CUTOFF
5.0e-6
&END
&SYSTEM
ANGSTROM
SYMMETRY
1
CHARGE
0.0
CELL
11.3 1.0 1.0 0.0 0.0 0.0
CUTOFF
25.0
&END
&ATOMS
*C_VDB_BLYP.psp NEWF FORMATTED
LMAX=P
1
[COORDINATES ]
*O_VDB_BLYP.psp NEWF FORMATTED
LMAX=P
48
[ COORDINATES ]
*H_VDB_BLYP.psp NEWF FORMATTED
LMAX=S
92
[ COORDINATES ]
&END
More information about the CPMD-list
mailing list