[gmx-users] step 0Segmentation fault: 11

[Justin Lemkul]

On 11/27/13 7:30 PM, Valentina wrote:
> I still cannot resolve this problem. Is there any way to find out what is
> causing Segmentation Fault? Normally Gromacs provides quite comprehendible
> errors, but this one.
>
> I have checked the charges against DFT.
>
> I also have the output of the system simulated with Discover. I am using
> that force field (obviously tabulated) and charges used there.
>
> The initial model is taken from Xray and optimised with DFT.
>
> Still md.log terminates with unrealistic T & P on 0th step:
>
>       /      Step           Time         Lambda
>                0        0.00000        0.00000
>
>
> Grid: 4 x 3 x 3 cells
>     Energies (kJ/mol)
>       Tab. Bonds    Tab. Angles        LJ (SR)   Coulomb (SR)      Potential
>      5.44943e+06    4.41001e+11    9.86202e+02   -2.42548e+05    4.41006e+11
>      Kinetic En.   Total Energy  Conserved En.    Temperature Pressure (bar)
>      1.97584e+16    1.97588e+16    1.97588e+16    2.24716e+15    1.99874e+16
> /
>
>
> BTW - I noticed that I get continuously get a 0 dimension in z direction if
> I do energy minimisation with plc = xy - is that right?
>
> My equil.mdp is following - may be some problem there?
> I do energy minimisation before equilibration.
>
> /integrator               = md
> nsteps                   = 100
> dt			 = 0.001
> ;comm_mode = ANGULAR
> nstenergy                = 1
> nstxout	                 = 1
> nstvout	                 = 1
> energygrps               = System
> ; cut-offs at 1nm
> rlist                    = 0.8
> nstlist                  = 10
> coulombtype              = user
> rcoulomb                 = 0.8
> vdw-type                 = user
> rvdw                     = 0.8
> pbc                      = xyz
> periodic_molecules       = yes
> ; set temperature to 300K
> tcoupl                   = V-rescale
> tc-grps                  = System
> tau-t                    = 1.0
> ref-t                    = 500
> ; and pressure to 1 bar
> pcoupl          = no
> compressibility  =  4.5e-5
> ; generate initial velocities
> gen-vel                  = no/
>
> I am running with 6-9 LJ, that I provide input to mdrun
>
>
>
>
> I do energy minimisation before equilibration.
> em.mdp is
> /integrator               = steep
> nsteps                   = 500000
> emtol                    = 1
> emstep                   = 0.001
> nstxout                  = 1
> nstenergy                = 1
> rlist                    = 0.8
> pbc                      = xyz
> periodic_molecules       = yes
> coulombtype              = user
> rcoulomb                 = 0.8
> vdw-type                 = user
> rvdw                     = 0.8
> constraints             = none/
>
>
> I get following output:
>
> /Steepest Descents:
>     Tolerance (Fmax)   =  1.00000e+00
>     Number of steps    =       500000
> Step=    2, Dmax= 5.0e-04 nm, Epot=  4.40635e+11 Fmax= 8.77355e+10, atom=
> 223
> Step=    4, Dmax= 3.0e-04 nm, Epot=  4.40403e+11 Fmax= 9.61163e+10, atom=
> 263
> Step=    6, Dmax= 1.8e-04 nm, Epot=  4.40248e+11 Fmax= 1.01812e+11, atom=
> 833
> Step=    8, Dmax= 1.1e-04 nm, Epot=  4.40181e+11 Fmax= 9.36260e+10, atom= 83
> Step=   10, Dmax= 6.5e-05 nm, Epot=  4.40142e+11 Fmax= 9.89456e+10, atom=
> 833
> Step=   12, Dmax= 3.9e-05 nm, Epot=  4.40115e+11 Fmax= 8.24990e+10, atom=
> 105
> Step=   14, Dmax= 2.3e-05 nm, Epot=  4.40094e+11 Fmax= 4.88140e+10, atom=
> 105
> Step=   19, Dmax= 1.7e-06 nm, Epot=  4.40098e+11 Fmax= 4.47573e+10, atom=
> 223
> Energy minimization has stopped, but the forces havenot converged to the
> requested precision Fmax < 1 (whichmay not be possible for your system). It
> stoppedbecause the algorithm tried to make a new step whose sizewas too
> small, or there was no change in the energy sincelast step. Either way, we
> regard the minimization asconverged to within the available machine
> precision,given your starting configuration and EM parameters.
>
> Double precision normally gives you higher accuracy, butthis is often not
> needed for preparing to run moleculardynamics.
> You might need to increase your constraint accuracy, or turn
> off constraints altogether (set constraints = none in mdp file)
>
> writing lowest energy coordinates.
>
> Back Off! I just backed up confout.gro to ./#confout.gro.4#
>
> Steepest Descents converged to machine precision in 20 steps,
> but did not reach the requested Fmax < 1.
> Potential Energy  =  4.4009390e+11
> Maximum force     =  4.8813974e+10 on atom 105
> Norm of force     =  5.4008571e+09
> /
>
> I cannot recollect how energy minimisation output should look like - is this
> right?
>

No, this outcome is terrible.  Energy should (almost) always be negative and the 
maximum force should be less than the target Fmax.  You have massive forces 
(10^10 - yikes!) on atom 105, indicating either severe problems with the 
configuration (which should be obvious upon visual inspection) or, again, your 
tabulated potential is wrong.

Mark's suggestion before is a good one - simplify the system into its most basic 
units, down to the level of two atoms and build up to verify your input.  From 
the output shown, the only logical explanations are (1) bad tables or (2) bad 
configuration.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441

==================================================