[Pw_forum] magic cell size in FPMD
Davide Ceresoli
ceresoli at sissa.it
Tue Nov 8 17:54:44 CET 2005
I'm running a simple calculation (1 atom in a cubic cell),
in serial, on a PC, and vary the cell size (celldm(1)).
celldm(1) = 18.897261 (10 angstrom) everything is fine!
celldm(1) = 18.0 crashes when writing restart to disk
celldm(1) = 18.8 crashes when writing restart to disk
celldm(1) = 19.0 crashes when writing restart to disk
celldm(1) = 22.676714 (20 angstrom) crashes when writing restart to disk
It's FPMD from CVS (end of september 2005), compiled with intel 8.1.
Debugger tells me it stops in Modules/mp_wave.f90:116
114 DO I = 1, ngwl
115 ! WRITE( stdout,*) 'MW ', ig_l2g(i), i
116 PWT( ig_l2g(i) ) = pw(i)
117 END DO
Luckily, up to now, I've always been using 10 angstroms!
Here is a sample input file:
&control
title = 'hydrogen atom'
prefix = 'h_sic'
calculation = 'cp'
restart_mode = 'from_scratch'
ndr = 50
ndw = 50
nstep = 100
iprint = 10
isave = 2
tstress = .true.
tprnfor = .true.
dt = 2.0
etot_conv_thr = 1.d-9
outdir = './output/'
pseudo_dir = './'
/
&system
ibrav = 1
celldm(1) = 18.897261
nat = 1
ntyp = 1
ecutwfc = 40.0
nbnd = 1
nspin = 2
nelup = 1
neldw = 0
nelec = 1
ecutwfc = 80.0
xc_type = 'PW'
sic = 'sic_mac'
force_pairing = .true.
/
&electrons
emass = 400
emass_cutoff = 2.5
orthogonalization = 'ortho'
ortho_eps = 5.d-8
ortho_max = 15
electron_dynamics = 'sd'
!electron_dynamics = 'damp'
!electron_damping = 0.2
electron_velocities = 'zero'
electron_temperature = 'not_controlled'
startingwfc = 'atomic'
/
&ions
ion_dynamics = 'none'
ion_radius(1) = 0.8d0
ion_velocities = 'zero'
ion_temperature = 'not_controlled'
sic_rloc = 1.0
/
&cell
cell_dynamics = 'none'
cell_velocities = 'zero'
press = 0.0
wmass = 70000.0
/
ATOMIC_SPECIES
H 1.0 H.blyp-vbc.UPF
K_POINTS gamma
ATOMIC_POSITIONS angstrom
H 0.0000 0.0000 0.0000 0 0 0 1
EOF
More information about the Pw_forum
mailing list