[gmx-users] Inconsistent shifts - is it something wrong with pbc

[Justin A. Lemkul]

ming ma wrote:
> Hi,
> 
> Thanks for looking at this email. Before I asked this question, I have 
> read most of the related topics in gromacs's mailing list, but still 
> can't solve it.
> When I was trying to do energy minimization of graphene with pbc, I got 
> several messages "there are # inconsistent shifts, check your topolgy."  
> After the mimization is finished, the energy in extremely large 
> (1.86e+05 kJ/mol), which is obviously wrong. To make sure that I haven't 
> done any stupid mistake, I have carried out energy minimization of the 
> same system using LAMMPS, and the results is reasonable (the energy is 
> -205.14 eV). The configuration of the system has also been checked by 
> visualizing it in VMD, and the PBC seems right.
> 
> Below is my mdp file,
> 
> ; energy minimization
> integrator               = l-bfgs
> nsteps                   = 5000
> constraints              = none
> emtol                    = 1.0
> nstcgsteep               = 10 ; do a steep every 10 steps of cg
> emstep                   = 0.01 ; used with steep
> nbfgscorr                = 10
> nstcomm                  = 0 ; frequecy of remove of com movement
> Tcoupl                   = no
> Pcoupl                   = no
> 
> ; potential and neigbor related
> coulombtype              = Cut-off
> rcoulomb                 = 1.2
> vdw-type                 = Switch
> rvdw-switch              = 0.8
> rvdw                     = 1.0
> ; neighbor list related
> rlist                    = 1.2
> nstlist                  = 10
> ns_type                  = grid
> 
> pbc                      = xyz
> ;output section
> nstxout                  = 0
> nstvout                  = 0
> nstfout                  = 0
> nstlog                   = 100
> nstenergy                = 100
> nstxtcout                = 0
> xtc_precision            = 1000
> xtc-grps                 =
> energygrps               =
> 
> And these are parts of the message I got during simulation,
> 
> There were 262 inconsistent shifts. Check your topology
> Will stop reporting inconsistent shifts
> 
> This is the results I got
> Low-Memory BFGS Minimizer converged to Fmax < 1 in 432 steps
> Potential Energy  =  1.85767014488657e+05
> 
> I also tried using the cg method instead of l-bfgs, after read the log 
> files carefully, I found these messages,
> 
> Initializing Domain Decomposition on 4 nodes
> Dynamic load balancing: no
> Will sort the charge groups at every domain (re)decomposition
> Initial maximum inter charge-group distances:
>     two-body bonded interactions: 14.880 nm, Morse, atoms 8401 8540
>   multi-body bonded interactions: 20.914 nm, G96Angle, atoms 8470 8540
> Minimum cell size due to bonded interactions: 23.006 nm
> 
> I checked the distances between the atoms it showed, obviously, this 
> large distance is caused by not having taken the pbc xyz into 
> consideration since the size of the simulation box is 15*15**6 nm. So 
> does it mean I haven't handle the pbc correctly? Since I notice in the 
> message posted before some of the 'inconsistent shifts' are also caused 
> by pbc problems.
> 

Exactly, the periodicity is not being handled correctly.  Presumably you're 
dealing with an infinite sheet (thus bonds extend across periodic boundaries), 
and thus you need to set "periodic_molecules = yes" in the .mdp file.

-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

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