VASP stops when computing dielectric tensor with hybrid fu
Posted: Sat Mar 14, 2015 9:44 pm
Hi,
I'm trying a calculation of the dielectric tensor with hybrid functional. I find that after the two first scf cycles the program stops. stdout ends with many repeated instances of these three lines:
PEAD_POLARIZATION_CALC: WARNING: reciprocal direction 3 spin channel 1
contribution from string nn not well clustered around <|S|>_av = ( 0.43626 0.03294 )
Im(ln[|S|/<|S|>_av])/2pi = yyy > 1/4
where nn is an integer (in my case up to 64) and yyy is very close to 0.5 or -0.5; these lines are followed by another two lines:
p_tot=( 0.232E+01 -0.862E+00 0.210E+01 )
48 total processes killed (some possibly by mpirun during cleanup)
The last lines in OUTCAR are:
------------------------ aborting loop because EDIFF is reached ----------------------------------------
CALCP: cpu time 103.17: real time 103.20
Ionic dipole moment: p[ion]=( 17.95341 30.31005 0.10297 ) electrons Angst
Spin resolved dipole moment: p[sp1]=( -0.63679 0.48079 -0.28541 ) electrons Angst
p[sp2]=( 2.95906 -1.34307 2.38727 ) electrons Angst
Total electronic dipole moment: p[elc]=( 2.32227 -0.86228 2.10186 ) electrons Angst
FORLOC: cpu time 0.03: real time 0.03
The INCAR file is:
SYSTEM = opt
NCORE = 1
KPAR=16
ALGO = Damped ; TIME = 0.5
ISMEAR = -5
PREC = Normal
LREAL=False
ENCUT=550
LASPH = .TRUE. ; ADDGRID = .TRUE. ; LMAXMIX = 4
ISPIN = 2
EDIFF=3E-05
ICHARG=1
LHFCALC = .TRUE.; AEXX = 0.24
PRECFOCK = Fast
NKRED=4
LCALCEPS=.TRUE. ; EFIELD PEAD = 0.002 0.002 0.002
IBRION = -1
NSW = 0
LWAVE=.FALSE
I'm using 6 nodes of an AMD machine (8 cores/node) connected with infiniband; program was compiled with ifort 11 and mkl libraries. PAW potentials are being used.
Any help on how to solve the problem?
Thanks in advance,
José C. Conesa
I'm trying a calculation of the dielectric tensor with hybrid functional. I find that after the two first scf cycles the program stops. stdout ends with many repeated instances of these three lines:
PEAD_POLARIZATION_CALC: WARNING: reciprocal direction 3 spin channel 1
contribution from string nn not well clustered around <|S|>_av = ( 0.43626 0.03294 )
Im(ln[|S|/<|S|>_av])/2pi = yyy > 1/4
where nn is an integer (in my case up to 64) and yyy is very close to 0.5 or -0.5; these lines are followed by another two lines:
p_tot=( 0.232E+01 -0.862E+00 0.210E+01 )
48 total processes killed (some possibly by mpirun during cleanup)
The last lines in OUTCAR are:
------------------------ aborting loop because EDIFF is reached ----------------------------------------
CALCP: cpu time 103.17: real time 103.20
Ionic dipole moment: p[ion]=( 17.95341 30.31005 0.10297 ) electrons Angst
Spin resolved dipole moment: p[sp1]=( -0.63679 0.48079 -0.28541 ) electrons Angst
p[sp2]=( 2.95906 -1.34307 2.38727 ) electrons Angst
Total electronic dipole moment: p[elc]=( 2.32227 -0.86228 2.10186 ) electrons Angst
FORLOC: cpu time 0.03: real time 0.03
The INCAR file is:
SYSTEM = opt
NCORE = 1
KPAR=16
ALGO = Damped ; TIME = 0.5
ISMEAR = -5
PREC = Normal
LREAL=False
ENCUT=550
LASPH = .TRUE. ; ADDGRID = .TRUE. ; LMAXMIX = 4
ISPIN = 2
EDIFF=3E-05
ICHARG=1
LHFCALC = .TRUE.; AEXX = 0.24
PRECFOCK = Fast
NKRED=4
LCALCEPS=.TRUE. ; EFIELD PEAD = 0.002 0.002 0.002
IBRION = -1
NSW = 0
LWAVE=.FALSE
I'm using 6 nodes of an AMD machine (8 cores/node) connected with infiniband; program was compiled with ifort 11 and mkl libraries. PAW potentials are being used.
Any help on how to solve the problem?
Thanks in advance,
José C. Conesa