dimer calculation error message

[kai]

Hey, there,
I am running a dimer in DFT-D and start to see some error message that i dont understand; any insight would be helpful.

Simply put, this is from the transition state of a set of pre-converged NEB calculation (0.05 eV/A) and the out info as follows:


running on 64 nodes
distr: one band on 1 nodes, 64 groups
vasp.5.2.11 18Jan11 gamma-only
POSCAR found type information on POSCAR Br C H Si
POSCAR found : 4 types and 266 ions
LDA part: xc-table for Pade appr. of Perdew
POSCAR, INCAR and KPOINTS ok, starting setup
WARNING: small aliasing (wrap around) errors must be expected
FFT: planning ...( 1 )
WAVECAR not read
WARNING: random wavefunctions but no delay for mixing, default for NELMDL
entering main loop
N E dE d eps ncg rms rms(c)
RMM: 1 0.846318867417E+04 0.84632E+04 -0.19116E+05 1536 0.625E+02
RMM: 2 0.227454914888E+04 -0.61886E+04 -0.68506E+04 1536 0.133E+02
RMM: 3 0.263644504128E+03 -0.20109E+04 -0.26792E+04 1536 0.767E+01
RMM: 4 -0.733653699343E+03 -0.99730E+03 -0.83142E+03 1536 0.541E+01
RMM: 5 -0.102731973103E+04 -0.29367E+03 -0.24030E+03 1536 0.327E+01
RMM: 6 -0.112173226229E+04 -0.94413E+02 -0.77604E+02 1536 0.203E+01
RMM: 7 -0.115222943451E+04 -0.30497E+02 -0.26860E+02 1536 0.127E+01
RMM: 8 -0.116337849528E+04 -0.11149E+02 -0.99333E+01 1536 0.814E+00
RMM: 9 -0.116981270645E+04 -0.64342E+01 -0.61171E+01 3471 0.533E+00
RMM: 10 -0.117066531561E+04 -0.85261E+00 -0.75403E+00 3831 0.169E+00
RMM: 11 -0.117069450101E+04 -0.29185E-01 -0.24718E-01 3427 0.259E-01
RMM: 12 -0.117069919209E+04 -0.46911E-02 -0.44742E-02 3824 0.106E-01 0.604E+01
RMM: 13 -0.126468826848E+04 -0.93989E+02 -0.13962E+02 3156 0.722E+00 0.455E+01
RMM: 14 -0.129319174491E+04 -0.28503E+02 -0.45440E+01 3125 0.357E+00 0.274E+01
RMM: 15 -0.128895999824E+04 0.42317E+01 -0.16400E+01 3158 0.230E+00 0.188E+01
RMM: 16 -0.128633353056E+04 0.26265E+01 -0.46000E+00 3148 0.126E+00 0.149E+01
RMM: 17 -0.128328412352E+04 0.30494E+01 -0.46221E+00 3103 0.132E+00 0.808E+00
RMM: 18 -0.128226946983E+04 0.10147E+01 -0.20775E+00 3136 0.916E-01 0.454E+00
RMM: 19 -0.128231351970E+04 -0.44050E-01 -0.62296E-01 3176 0.476E-01 0.311E+00
RMM: 20 -0.128222600815E+04 0.87512E-01 -0.10687E-01 3234 0.195E-01 0.252E+00
RMM: 21 -0.128219244755E+04 0.33561E-01 -0.14977E-01 3131 0.186E-01 0.174E+00
RMM: 22 -0.128215264005E+04 0.39808E-01 -0.13068E-01 3164 0.195E-01 0.121E+00
RMM: 23 -0.128214086923E+04 0.11771E-01 -0.55992E-02 3148 0.123E-01 0.118E+00
RMM: 24 -0.128212563845E+04 0.15231E-01 -0.22227E-02 3151 0.817E-02 0.538E-01
RMM: 25 -0.128212351009E+04 0.21284E-02 -0.14917E-02 3109 0.701E-02 0.458E-01
RMM: 26 -0.128212315259E+04 0.35750E-03 -0.47596E-03 3200 0.403E-02 0.217E-01
RMM: 27 -0.128212379325E+04 -0.64066E-03 -0.38156E-03 3159 0.318E-02 0.201E-01
RMM: 28 -0.128212429014E+04 -0.49688E-03 -0.18909E-03 3167 0.230E-02 0.173E-01
RMM: 29 -0.128212471804E+04 -0.42790E-03 -0.10188E-03 3179 0.182E-02 0.167E-01
RMM: 30 -0.128212464756E+04 0.70480E-04 -0.70695E-04 3173 0.136E-02 0.508E-02
RMM: 31 -0.128212470349E+04 -0.55930E-04 -0.26894E-04 3213 0.900E-03 0.460E-02
RMM: 32 -0.128212471976E+04 -0.16268E-04 -0.10400E-04 3193 0.584E-03 0.228E-02
RMM: 33 -0.128212472544E+04 -0.56863E-05 -0.71247E-05 3168 0.479E-03 0.237E-02
RMM: 34 -0.128212472561E+04 -0.16262E-06 -0.98039E-06 3127 0.206E-03 0.128E-02
RMM: 35 -0.128212472979E+04 -0.41814E-05 -0.12791E-05 3128 0.206E-03 0.645E-03
RMM: 36 -0.128212473062E+04 -0.83116E-06 -0.28881E-06 3138 0.106E-03 0.495E-03
RMM: 37 -0.128212473035E+04 0.27187E-06 -0.12848E-06 3158 0.679E-04 0.316E-03

DSYGV : 2538-2115
The matrix (ARG NO. 4) is not positive definite.
The leading minor of order (3) has a nonpositive determinant.

DSYGV : 2538-2604
Execution terminating due to error count for error number 2115.

DSYGV : 2538-2605
Message summary: 2115 - 1


############
INCAR

IALGO = 48 # use RMM-DIIS algorithm for electrons
MAXMIX = 50 # store 50 steps in charge mixer before reseting (even if ions move

LREAL = A #.FALSE. # A = real space (big system), .false. use reciprocal space (small systems). Manual recommends A for mixed ions or more than 20 atoms
NSIM = 4
ISPIN = 2 # spin polarized calculation

#ion convergence criteria are here
NSW = 999
IBRION = 3
POTIM = 0
EDIFFG = -0.02
ISMEAR = 0 # recommended for semiconductor
SIGMA = 0.05 # recommended for semiconductor
LVDW=.TRUE.

#electronic convergence
EDIFF = 1E-8 # 1E-4 default
NELMIN = 8 # switches for fast convergence - see manual page 48
NELM = 100 # maximum of electronic steps
ENCUT = 450

AMIX = 0.1 #set amix
BMIX = 0.0001 #almost zero in initial mixer


PREC = Medium # control NGX, NGY, NGZ and energy cutoff

LWAVE = .FALSE. # not write wavecar
LCHARG = .FALSE. # not write chargecar

#NEB or Dimer switched
ICHAIN = 2 # 0 is NEB; 2 is dimer
IOPT=7

[graeme]

Can you check to see if this is related to the dimer calculation? To do this, you can comment the INCAR lines:

#ICHAIN = 2 # 0 is NEB; 2 is dimer
#IOPT=7

My expectation is that you are having problems with convergence of the electronic structure and that it is unrelated to doing a dimer calculation. If you get the same message with those lines commented, it is not an issue with our code.

If this is the case, you can try using a different (or less aggressive) IALGO choice.

[kai]

you are right....it is the electronic convergence...thx

[kai]

further update, though i dont understand

it seems to be associate with the gamma only version vasp; the regular one runs nicely. (the version we are running is 5.2.11)

my labmate had similar experience when running NEB using ibrion=3 (with gamma only version), but fine with ibrion=1(again gamma only)

a trick we used to get around the problem is to set

EDIFF = 1E-7 #(slightly larger than what we want, 1-e-8)
NELMIN = 10 # but define a large min number here to force it gets down to 1e-8 after first couple of force call

it seems ok