Pressure convergence issue in NPT molecular dynamics

[fanghz]

I have tried a few NPT MD simulations and found one important issue that the external pressures in OUTCAR file never converge to the target pressure (supposed to be the value of PSTRESS ?), especially at high temperatures (I did found it converges to PSTRESS at room temperature). Below is an example of my INCAR setting. Please let me know if you have any comments, thanks in advance!

SYSTEM = LaLiTiO3

# Settings for parallel version (ignored in serial run)
LPLANE = .TRUE. ! distribute each wave over processors
NCORE = 8 ! number of cores per node
NSIM = 4
LSCALU = .FALSE.
LSCALAPACK = .FALSE.

# Electronic Relaxation 1
PREC = Normal
ENCUT = 520
LREAL = A
ALGO = Fast
EDIFF = 1e-4
NBANDS = 704
NELMIN = 4
NELM = 40

# Mixing Tags
# AMIX = 0.5
# BMIX = 4.0
MAXMIX = 45

# Ionic Relaxation
IBRION = 0
ISIF = 3
MDALGO = 3
PSTRESS = 0.001
LANGEVIN_GAMMA = 69.5 3.5 23.9 8.0
LANGEVIN_GAMMA_L = 50
PMASS = 10
NSW = 20000
SMASS = 0
POTIM = 1.5
TEBEG = 2723
TEEND = 2723
NBLOCK = 1; KBLOCK = 100

# Symmetry settings
ISYM = 0

# DOS related values
ISMEAR = -1
SIGMA = 0.2347

# output
LCHARG = F
LWAVE = F

[admin]

The pressure written in OUTCAR does not include the contribution
from the kinetic stress. The procedure how to calculate the correct value is described here:
http://cms.mpi.univie.ac.at/vasp/vasp/P ... ostat.html