vasp broke down without any error information

[lgxyz]

Hi, everyone, recently my NEB jobs boke dowm without any error information. The following is some of the OUTCAR:

..........
----------------------------------------- Iteration 5( 66) ---------------------------------------


POTLOK: VPU time 0.65: CPU time 0.65
SETDIJ: VPU time 0.08: CPU time 0.08
EDDIAG: VPU time 17.26: CPU time 17.26
RMM-DIIS: VPU time 96.13: CPU time 96.14
ORTHCH: VPU time 3.62: CPU time 3.62
DOS : VPU time 0.00: CPU time 0.00
CHARGE: VPU time 4.51: CPU time 4.51
MIXING: VPU time 0.33: CPU time 0.34
------------------------------------------
LOOP: VPU time 122.61: CPU time 122.63

eigenvalue-minimisations : 3809
total energy-change (2. order) :-0.3890312E+00 ( 0.7830774E+01)
number of electron 384.0000034 magnetization
augmentation part 116.1809099 magnetization

Broyden mixing:
rms(total) = 0.11602E+01 rms(broyden)= 0.11602E+01
rms(prec ) = 0.12842E+01
weight for this iteration 100.00

eigenvalues of (default mixing * dielectric matrix)
average eigenvalue GAMMA= 0.4143
1.9520 1.9520 1.8790 1.0952 0.8751 0.8751 0.7934 0.6029 0.6029 0.4272
0.4272 0.3528 0.3528 0.2095 0.2095 0.3444 0.2604 0.2604 0.3141 0.3141
0.2541 0.2541 0.2670 0.1201 0.1201 0.1841 0.1841 0.0840 0.0840 0.1801
0.1604 0.1604 0.0256 0.0278 0.0278 0.0214 0.0583 0.0583 0.0995 0.0995

Free energy of the ion-electron system (eV)
---------------------------------------------------
alpha Z PSCENC = 2373.50892359
Ewald energy TEWEN = 19017.49122010
-1/2 Hartree DENC = -47915.26605992
-V(xc)+E(xc) XCENC = 215.63984054
PAW double counting = 28693.51649564 -27816.61238720
entropy T*S EENTRO = -0.18119239
eigenvalues EBANDS = -794.91222192
atomic energy EATOM = 26971.31666720
---------------------------------------------------
free energy TOTEN = 744.50128564 eV

energy without entropy = 744.68247803 energy(sigma->0) = 744.59188183


--------------------------------------------------------------------------------------------------------




----------------------------------------- Iteration 5( 67) ---------------------------------------


POTLOK: VPU time 0.65: CPU time 0.65
SETDIJ: VPU time 0.08: CPU time 0.08
EDDIAG: VPU time 17.25: CPU time 17.25

Stop just here, no error warning. Anybody has the same experience? Please give me some suggestions. Thanks so much!

[forsdan]

Well, first of all: have you check the std.out and the std.err files? Vasp, mpi and system errors are written there.

Secondly, please note that your energies are positive, so you are in a very unfavorable geometry. This might cause the electronic relaxation to break. Such an error should usually show up in the std.out in that case. Please check your geometry in a viewer to be sure that everything is set up correctly.


Cheers,
/Dan

<span class='smallblacktext'>[ Edited Sun Sep 12 2010, 04:05PM ]</span>

[lgxyz]

Thanks so much!
Where are the std.out and std.err files?
Before NEB jobs run, the two endpoints (the reactant and product) have been optimized. The images (intermediates)are produced by the script nebmake.pl, so how can I do some about geometries? Thank you!

[forsdan]

With std.out and std.err I refer to the system output, which is usually written to the display if you run interactively. The output can be written to different filenames depending on your system/cluster/scheduler configuration. Ask your system admin if you don't know where the output is written.

If you use an interpolation script, you must be very careful when it comes to atoms crossing the unit cell boundaries. If one atom has the coordinate 0.1 (0.9) in the starting (final) state, the interpolation will be 0.1--> 0.9 and not 0.1 --> -0.1. Therefore the atom will transverse your cell. Based on the high positive energies, this is most likely the case for you. Please make sure all intermediate images are what you expect them to be.

Cheers,
/Dan

[lgxyz]

Yes, it is. The geometries of the images interpolated by the script are not reasonable. If we do not use the script, is there any other methods to create images? Many many many thanks!