ERROR FEXCF

[lusommmg]

Hi, I get a problem when I calculate a system including Lanthnide with PAW-gga potantials using a supercell with 24 atoms,the job failed,and there is a erro in job.sh.o* file, it is like this :
------------------------------
3 F= -.17834534E+03 E0= -.17837361E+03 d E =-.438535E+00 mag= 0.0000
curvature: -0.41 expect dE=-0.102E+00 dE for cont linesearch -0.211E-03
trial: gam= 0.20797 g(F)= 0.156E+00 g(S)= 0.918E-01 ort = 0.234E-01 (trialstep = 0.965E+00)
search vector abs. value= 0.304E+00
bond charge predicted
N E dE d eps ncg rms rms(c)
DAV: 1 -0.177151336089E+03 -0.17715E+03 -0.45509E+01 76544 0.536E+01 0.353E+00
DAV: 2 -0.184084842951E+03 -0.69335E+01 -0.39853E+01123592 0.267E+01 0.160E+01
DAV: 3 -0.204290146194E+03 -0.20205E+02 -0.88469E+01112472 0.457E+01 0.447E+01
DAV: 4 -0.610396798466E+03 -0.40611E+03 -0.78805E+02120688 0.918E+01 0.113E+02
DAV: 5 -0.197575592731E+03 0.41282E+03 -0.43954E+02119528 0.793E+01 0.566E+01
DAV: 6 -0.215004486363E+03 -0.17429E+02 -0.15377E+02128280 0.294E+01 0.663E+01
DAV: 7 -0.203627472189E+03 0.11377E+02 -0.28835E+01125944 0.144E+01 0.477E+01
DAV: 8 -0.187145346858E+03 0.16482E+02 -0.35952E+01120240 0.237E+01 0.226E+01
DAV: 9 -0.189436972185E+03 -0.22916E+01 -0.36710E+00124624 0.918E+00 0.263E+01
DAV: 10 -0.193273210766E+03 -0.38362E+01 -0.11566E+01120352 0.116E+01 0.331E+01
DAV: 11 -0.184160703321E+03 0.91125E+01 -0.23725E+01118016 0.211E+01 0.194E+01
DAV: 12 -0.184031434995E+03 0.12927E+00 -0.24364E+00129920 0.588E+00 0.186E+01
DAV: 13 -0.229554442793E+03 -0.45523E+02 -0.79129E+01119496 0.337E+01 0.644E+01
DAV: 14 -0.131150842588E+06 -0.13092E+06 -0.12541E+07113608 0.399E+05 0.220E+02
DAV: 15 -0.568883159916E+05 0.74263E+05 -0.25495E+05136496 0.211E+03 0.291E+02
ERROR FEXCF: supplied exchange-correlation table
is too small, maximal index : 5459
----------------------------------------------------------
but there is no error at all when the supercell including 16 atoms, can anybody help me explain it and fix it out ? thank you very much!
the INCAR file as follows:
General
SYSTEM = hcp 24.1 optimization
PREC = High
ISTART = 0
ICHARG = 0

Electronic minimization
#ENCUT = 400
#ENAUG = 613.6
ISPIN = 2
ISIF = 3
NELM = 50
NELMIN = 4
NELMDL = -12
EDIFF = 1E-4
MAXMIX = 80
LREAL = .FALSE.
ADDGRID = .TRUE.
#NBANDS = 32
LWAVE = .FALSE.
LCHARG = .FALSE.

Ionic relaxation
IALGO = 38
NSW = 40
IBRION = 2
POTIM = 0.5
#BMIX = 0.8
EDIFFG = 1E-3

DOS related parameters
ISMEAR = 1
SIGMA = 0.2

[gaurav]

I am Also facing the same problem. dont know how to solve it .p lease help

[admin]

this error usually indicates that the ionic configuration gives an unreaonable charge density, therefore my first guess is that your ionic relaxation is unphysical.
please have a look at your results of steps 1-3: does the total energy decrease? are the predicted geometries and interatomic distances reasonable? (If you insist on volume relaxation: does the external pressure converge?)
1) please consider whether it is reasonable to relax the cell volume and shape of a defective supercell (especially as it only contains 24 atoms!, i e the macroscopic defect concentration would be about 4%!)
2) EDIFF = 1e-4 is definitely not accurate enough if you want to optimize the structure based on the force tensors, this requires better electronic convergence (at least 1e-6)
3) POTIM may be too large, your first relaxation step might have lead to an unresonable 2nd cell configuration