ERROR FEXCP: supplied Exchange-correletion table

[gus.hart]

The error has occurred with a handful of binary/ternary structures, ideal fcc supercells, with atoms on ideal lattice sites (Ni, Ti, and Hf). The volumes are slightly larger than the fully relaxed case (slightly negative pressures). The first ionic step runs fine, but at the beginning of the second, the error is triggered.

It must be related to symmetry because if I distort the unit cell slightly, the error goes away (but this is an inefficient solution as many more kpoints are then needed...).

Code: Select all

MLIP output to VASP                     
   1.00000000000000     
     1.6569974008899999    1.6569974008899999   -1.6569974008899999
    -1.6569974008899999    4.9709922026699997    1.6569974008899999
     1.6569974008899999    1.6569974008899999    4.9709922026699997
   Hf   Ni
     1     3
Direct
  0.0000000000000000  0.0000000000000000  0.0000000000000000
  0.5000000000000000  0.5000000000000000  0.0000000000000000
  0.0000000000000000  0.5000000000000000  0.5000000000000000
  0.5000000000000000  0.0000000000000000  0.5000000000000000

And my INCAR:

Code: Select all

EDIFFG=1e-6
NSW=30    
PREC=Accurate
ENCUT=368    
ISMEAR=2  
SIGMA=.01 
NCORE=2   
IBRION=1  
POTIM=1   
ISIF=3   

And this is the output:

Code: Select all

 running on    2 total cores
 distrk:  each k-point on    2 cores,    1 groups
 distr:  one band on    2 cores,    1 groups
 using from now: INCAR     
 vasp.5.4.4.18Apr17-6-g9f103f2a35 (build Nov 06 2019 09:29:44) complex          
  
 POSCAR found type information on POSCAR  Hf Ni
 POSCAR found :  2 types and       4 ions
 scaLAPACK will be used
 LDA part: xc-table for Pade appr. of Perdew
 POSCAR, INCAR and KPOINTS ok, starting setup
 FFT: planning ...
 WAVECAR not read
 entering main loop
       N       E                     dE             d eps       ncg     rms          rms(c)
DAV:   1     0.513267535195E+03    0.51327E+03   -0.33640E+04 53504   0.162E+03
DAV:   2     0.383616667962E+01   -0.50943E+03   -0.50070E+03 41624   0.541E+02
DAV:   3    -0.260244998308E+02   -0.29861E+02   -0.28259E+02 52710   0.123E+02
DAV:   4    -0.277637185265E+02   -0.17392E+01   -0.16493E+01 49250   0.237E+01
DAV:   5    -0.278418422729E+02   -0.78124E-01   -0.71411E-01 50104   0.440E+00    0.529E+01
DAV:   6    -0.397461630533E+02   -0.11904E+02   -0.10367E+02 50790   0.100E+02    0.421E+01
DAV:   7    -0.250291518641E+02    0.14717E+02   -0.76993E+01 50870   0.732E+01    0.212E+01
DAV:   8    -0.246432029168E+02    0.38595E+00   -0.33468E+00 47366   0.122E+01    0.870E+00
DAV:   9    -0.246574552460E+02   -0.14252E-01   -0.18537E-01 55642   0.241E+00    0.865E+00
DAV:  10    -0.246606150554E+02   -0.31598E-02   -0.24144E-01 56844   0.378E+00    0.285E+00
DAV:  11    -0.246514367564E+02    0.91783E-02   -0.11175E-02 43710   0.687E-01    0.215E+00
DAV:  12    -0.246310467136E+02    0.20390E-01   -0.59750E-02 58002   0.111E+00    0.579E-01
DAV:  13    -0.246320279035E+02   -0.98119E-03   -0.46513E-03 53366   0.372E-01    0.490E-01
DAV:  14    -0.246314370311E+02    0.59087E-03   -0.77332E-04 53298   0.191E-01    0.411E-01
DAV:  15    -0.246313580610E+02    0.78970E-04   -0.90609E-05 54460   0.480E-02
   1 F= -.24631358E+02 E0= -.24631360E+02  d E =-.246314E+02
 BRION: g(F)=  0.756E-57 g(S)=  0.115E+02
 bond charge predicted
       N       E                     dE             d eps       ncg     rms          rms(c)
 ERROR FEXCP: supplied Exchange-correletion table
    is too small, maximal index :        3866

[henrique_miranda]

I am not able to reproduce this issue locally.
Could you attach the OUTCAR as well?
In particular, I want to check which POTCARS you are using as well as the section 'exchange correlation table'.

[gus.hart]

Attached are the requested files

[henrique_miranda]

Ah, ok now I can reproduce the issue.
I tried with different pseudopotentials: PAW_PBE Hf and PAW_PBE Ni.
You are using: PAW_PBE Hf_pv and PAW_PBE Ni_pv.

Anyway, this issue was already reported in a previous thread:
https://www.vasp.at/forum/viewtopic.php?f=4&t=1491

In this particular case, I also have to point out that when performing ionic relaxations where the cell volume is allowed to change it is recommended to use a larger cutoff than the one specified in the POTCAR:
https://www.vasp.at/wiki/index.php/ISIF
(see note at the very end)

Setting ENCUT=400 seems to solve the issue.
Note however that this might be just because the stress is computed more accurately in the first step and the volume change is not so large.
If you find a case where increasing the cutoff does not solve the issue please report back to us.

Hope this helps!