[gmx-users] Gromacs DFTB problem

[Gábor Balogh]

Dear Gromacs users,
I have compiled a DFTB implementation for GROMACS:
http://cbp.cfn.kit.edu/joomla/index.php/downloads/18-gromacs-with-qm-mm-using-dftb3

The system consists of the catalytic domain of a protein in explicit
solvent (TIP3P). Long-range electrostatics is calculated using the PME
scheme. Conjugate-gradient energy minimization runs fine, but NVT
equilibration terminates after 16-44 ps (depending on PME options and
velocity generation) with the following error message:

DSYGVD: ier = 214
EXITING!

MD options:

define          = -DPOSRES_SIM
;
integrator      = md
; 1 ns
nsteps          = 1000000
dt              = 0.001
; Output control
nstxout         = 20000
nstvout         = 20000
nstenergy       = 10000
nstlog          = 2000
nstxtcout       = 10000
xtc-grps        = Protein
;
continuation    = no
constraint_algorithm = lincs
constraints     = all-bonds
;
ns_type         = grid
nstlist         = 10
rlist           = 1.0
rcoulomb        = 1.0
rvdw            = 1.0
;
coulombtype     = PME
pme_order       = 4
fourierspacing  = 0.12
;
tcoupl          = V-rescale
tc-grps         = Protein_Link   Water_and_ions
tau_t           = 0.1   0.1
ref_t           = 310   310
;
pcoupl          = no
; Periodic boundary conditions
pbc             = xyz
;
DispCorr        = EnerPres
;
gen_vel         = yes
gen_temp        = 310
gen_seed        = -1
;
cutoff-scheme   = group
; dftb
QMMM    = yes
QMMM-grps       = QMsystem      ; needs to be specified in the index file
QMMMscheme      = normal
QMmethod        = RHF   ; required but ignored
QMbasis         = STO-3G        ; required but ignored
QMcharge        = -1    ; an integer
MMchargescalefactor     = 1.    ; or smaller
; DFTB
QMdftbsccmode   = 3     ; for DFTB3, or 2 for DFTB2 (a.k.a. SCC-DFTB)
QMdftb-telec    = 310.  ; "electronic temperature" for the Fermi
distribution
QMdftb-slko-path        = /scratch1/BG/gro/slko_gmx/    ; path to the
parameter files including the trailing slash
QMdftb-slko-separator   =       ; possibly the character between element
names
QMdftb-slko-lowercase           = yes   ; or no, in the file names ("Ca"
vs. "ca")
QMdftb-slko-suffix      = -c.spl        ; of the file names
QMdftb-partial-pme      = 0     ; accelerated PME calculations -
recommended (0 to switch off)
QMdftb-dispersion       = 0     ; for D3, or 0 for no dispersion
QMdftb-cdko     = 0     ; currently not implemented
QMdftb-mmhub-inf        = 1     ; currently not implemented

Last lines of Gromacs-DFTB output:
STEP 24856


The number of MM atoms is 15355
MM+ETC   TIME:     24791883
EWALD-SR START
DFTB SR  TIME:      1860179
EWALD-SR END
DFTB PRE TIME:      2965810
SCC ITER TIME:     68803390
DFTB PME TIME:      4833495
DFTB DIA TIME:     12378935
SCC ITER TIME:     20096442
DFTB PME TIME:      4796196
DFTB DIA TIME:     12375332
SCC ITER TIME:     20050119
DFTB PME TIME:      4764749
DFTB DIA TIME:     12401352
SCC ITER TIME:     20061356
DFTB PME TIME:      4765373
DFTB DIA TIME:      7992655

DSYGVD: ier = 214
EXITING!

Could somebody help me to find out why the simulation is crashing?
Thanks in advance.

Best regards,
Gabor