Lattice-Dependant SIGSEV

Problems running VASP: crashes, internal errors, "wrong" results.


Moderators: Global Moderator, Moderator

Post Reply
Message
Author
doylead
Newbie
Newbie
Posts: 1
Joined: Wed Mar 16, 2016 3:00 pm

Lattice-Dependant SIGSEV

#1 Post by doylead » Fri Mar 18, 2016 12:13 am

I keep finding my way back to the erorr:

Code: Select all

forrtl: severe (174): SIGSEGV, segmentation fault occurred
I'm not sure what additional information would be helpful - there isn't a whole lot more information in other files (I'm getting exit code 44544 when using the ASE/VASP interface, OSZICAR/OUTCAR show no signs of job termination - they just don't move past initialization). If there are any tests I should run, or additional flags to add, I'd be happy to get back with more.

This has happened in two different cases, and I can't tell if they're related:
  • 1) When optimizing a bulk crystal, trials with some lattice constants crash before getting started
    2) When attempting to optimize a supercell (to try different magnetic orderings)
I've turned symmetry off (ISYM=0), and moving the atoms out of higher-symmetry positions. I've also tried removing atoms rom the supercell (down to one atom in the most extreme case) to try to rule out memory limitations.

Does anyone have an idea what may be happening here, or suggestions for what I should try?

I'll post an example of each INCAR below to give a full list of the parameters I'm using. For the first case:

Code: Select all

 ENCUT = 600.000000
 AMIX_MAG = 0.800000
 BMIX = 0.000100
 BMIX_MAG = 0.000100
 AMIX = 0.200000
 SIGMA = 0.050000
 EDIFF = 1.00e-04
 EDIFFG = -5.00e-02
 PREC = Normal
 GGA = PE
 ALGO = fast
 LREAL = .FALSE.
 ISYM = 0
 ISPIN = 2
 ISMEAR = 0
 KPAR = 4
 NELM = 250
 NEDOS = 2001
 LORBIT = 11
 NSW = 2000
 LDAUPRINT = 2
 INIMIX = 0
 IBRION = 2
 LDAUTYPE = 2
 LMAXMIX = 4
 ICHARG = 1
 LASPH = .TRUE.
 LDAU = .TRUE.
 LVTOT = .FALSE.
 LDAUL = 2 -1 -1
 LDAUU = 6.450 0.000 0.000
 LDAUJ = 0.000 0.000 0.000
 MAGMOM = 4*1.5000 12*0.0000
For the second case:

Code: Select all

 ENCUT = 600.000000
 AMIX_MAG = 0.800000
 BMIX = 0.000100
 BMIX_MAG = 0.000100
 AMIX = 0.200000
 SIGMA = 0.050000
 EDIFF = 1.00e-04
 EDIFFG = -5.00e-02
 PREC = Normal
 GGA = PE
 ALGO = fast
 LREAL = .FALSE.
 ISYM = 0
 ISPIN = 2
 ISMEAR = 0
 KPAR = 1
 NELM = 250
 NEDOS = 2001
 LORBIT = 11
 NSW = 2000
 LDAUPRINT = 2
 INIMIX = 0
 IBRION = 2
 LMAXMIX = 4
 ICHARG = 1
 LASPH = .TRUE.
 LDAU = .TRUE.
 LVTOT = .FALSE.
 LDAUL = 2
 LDAUU = 3.300
 LDAUJ = 0.000
 MAGMOM = 1*3.0000

admin
Administrator
Administrator
Posts: 2921
Joined: Tue Aug 03, 2004 8:18 am
License Nr.: 458

Re: Lattice-Dependant SIGSEV

#2 Post by admin » Tue Apr 26, 2016 9:41 am

please check whether for the runs that crash the available memory is just too little
(remind that the number of basis funcions varies with the cell dimensions)

Post Reply