[gmx-users] Segmentation fault - Implicit solvent

[Justin A. Lemkul]

Steven Neumann wrote:
> 
> 
> On Thu, Mar 1, 2012 at 5:43 PM, Justin A. Lemkul <jalemkul at vt.edu 
> <mailto:jalemkul at vt.edu>> wrote:
> 
> 
> 
>     Steven Neumann wrote:
> 
> 
> 
>         On Thu, Mar 1, 2012 at 3:58 PM, Steven Neumann
>         <s.neumann08 at gmail.com <mailto:s.neumann08 at gmail.com>
>         <mailto:s.neumann08 at gmail.com <mailto:s.neumann08 at gmail.com>>__>
>         wrote:
> 
> 
> 
>            On Thu, Mar 1, 2012 at 3:32 PM, Steven Neumann
>            <s.neumann08 at gmail.com <mailto:s.neumann08 at gmail.com>
>         <mailto:s.neumann08 at gmail.com <mailto:s.neumann08 at gmail.com>>__>
>         wrote:
> 
> 
> 
>                On Thu, Mar 1, 2012 at 2:55 PM, Justin A. Lemkul
>                <jalemkul at vt.edu <mailto:jalemkul at vt.edu>
>         <mailto:jalemkul at vt.edu <mailto:jalemkul at vt.edu>>> wrote:
> 
> 
> 
>                    Steven Neumann wrote:
> 
>                        Dear Gmx Users,
>                         I am trying to run nvt simulation (equilibration) of
>                        the protein in implicit solvent. My mdp:
>                                         integrator = md ; leap-frog
>         integrator
> 
>                        nsteps = 1000000 ; 0.0005 * 1000000 = 0.5 ns
> 
>                        dt = 0.0005 ; 0.5 fs
> 
>                        ; Output control
> 
>                        nstxout = 10000
> 
>                        nstxtcout = 10000 ; xtc compressed trajectory output
>                        every 2 ps
> 
>                        nstenergy = 10000 ; save energies every 2 ps
> 
>                        nstlog = 10000 ; update log file every 2 ps
> 
>                        ; Bond parameters
> 
>                        continuation = no
> 
>                        constraints = none
> 
>                        ; Neighborsearching
> 
>                        ns_type = simple ; search neighboring grid cells
> 
>                        nstlist = 0 ; 10 fs
> 
>                        rlist = 0 ; short-range neighborlist cutoff (in nm)
> 
>                        ; Infinite box with no cutoffs
> 
>                        pbc = no
> 
>                        rcoulomb = 0 ; short-range electrostatic cutoff
>         (in nm)
> 
>                        coulombtype = cut-off
> 
>                        vdwtype = cut-off
> 
>                        rvdw = 0 ; short-range van der Waals cutoff (in nm)
> 
>                        epsilon_rf = 0
> 
>                        comm_mode = angular ; remove angular and com motions
> 
>                        ; implicit solvent
> 
>                        implicit_solvent = GBSA
> 
>                        gb_algorithm = OBC
> 
>                        gb_epsilon_solvent = 80.0
> 
>                        sa_surface_tension = 2.25936
> 
>                        rgbradii = 0
> 
>                        sa_algorithm = Ace-approximation
> 
>                        nstgbradii = 1
> 
>                        ; Temperature coupling is on
> 
>                        Tcoupl = v-rescale
> 
>                        tau_t = 0.1
> 
>                        tc_grps = system
> 
>                        ref_t = 298
> 
>                        ; Velocity generation
> 
>                        gen_vel = yes ; Velocity generation is on
> 
>                        gen_temp = 298.0
> 
>                        gen_seed = -1
> 
>                        Then after grompp I am trying to run the
>         simulation on
>                        the cluster:
> 
>                        mdrun -pd -deffnm nvt500ps
> 
>                        My log file:
> 
>                        Back Off! I just backed up nvt500ps.log to
>                        ./#nvt500ps.log.1#
> 
>                        Reading file nvt500ps.tpr, VERSION 4.5.4 (single
>         precision)
> 
>                        Starting 8 threads
> 
>                        Back Off! I just backed up nvt500ps.trr to
>                        ./#nvt500ps.trr.1#
> 
>                        Back Off! I just backed up nvt500ps.xtc to
>                        ./#nvt500ps.xtc.1#
> 
>                        Back Off! I just backed up nvt500ps.edr to
>                        ./#nvt500ps.edr.1#
> 
>                        starting mdrun 'Protein'
> 
>                        1000000 steps, 500.0 ps.
> 
>                        Segmentation fault
> 
>                                         Do you have any clue what is
>         happening?
> 
> 
>                    Try running in serial or with a maximum of 2 threads.
>          Your
>                    problem could be related to
>                    http://redmine.gromacs.org/____issues/777
>         <http://redmine.gromacs.org/__issues/777>
>                    <http://redmine.gromacs.org/__issues/777
>         <http://redmine.gromacs.org/issues/777>>.  You will need to
> 
>                    upgrade to 4.5.5 (serial should work on 4.5.4).
> 
>                    -Justin
> 
>                         Thank you. What do you mean by running in
>         serial? Well... with 2
>                threads it does not make sense to use implicit solvent. Will
>                4.5.5 resolve this problem?
>                
>                 Indeed, it works in serial. Will version 4.5.5 resolve
>         it run it on
>            e.g. 12 nodes?
>            
>          Sorry, the same problem :
>          
> 
> 
>     I think, as stated in the redmine issue cited before, the limit is 2
>     threads/processors.
> 
>     -Justin
> 
> 
> 
> I tried on one thread and on two threads and the error is still the 
> same. Without particle decomposition the problem remains. Any suggestions?
> 

Then the problem lies outside of the algorithms.  Your system has become 
instantly unstable, thus suggesting that you have not sufficiently minimized or 
equilibrated the structure.

-Justin

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

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

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