[gmx-users] How to increase the ratio of cell size to constrain length per error message

[Mark Abraham]

On 25/04/2012 6:28 AM, PAVAN PAYGHAN wrote:
>
> Dear Gromacs Users,
>
> I am using gromacs version is 4.5.3.and running my jobs on single node 
> with 8 cores.
>
> I have two different systems which contain about 425000 atoms (protein 
> + Lipid +SOL) one with bound ligand
>
> and another one unbound protein.I have successfully reached up to 
> NPT equilibration run step,
>

It is a poor idea to do equilibration with Parinello-Rahman, which is 
unstable when the system is not already close to equilibrium. For some 
reason people still do it, despite at least a post per week on this list 
suggesting against it, and a warning in the manual. Use Berendsen to fix 
the density, then equilibrate further with P-R to get the right ensemble.

> now I want to continue the same for production run. Without ligand I 
> am able to run successfully But the same system with ligand is 
> crashing with following error-
>
> D D cell  1 0 0 could only obtain 1520 of the 1521 atoms that are
>
> are connected via constraints from the neighbouring cells
>
> This probably means your constraint length are too long
>
> compared to the domain decomposition cell size.
>
> Decrease the number of domain decomposition grid cells or lincs_order.
>

I'd rather expect your system was blowing up because of the above issue. 
Perhaps the suggestion in the error message is not as complete as could 
be desired - you have so many atoms per processor that the constraint 
length would normally be tiny with respect to the cell size. So I think 
the things you have tried below are rearranging the deck chairs on the 
Titanic.

Mark

> Accordingly following the suggestions given in the error I tried to 
> solve it with
>
> Following log file and changed,
>
> 1.1. -rcon
>
>         Estimated maximum distance required for p-lincs was 0.877 thus 
> I increased it to 0.900
>
>   then it thrown another error.
>
>   The initial cell size (0.877) is smaller than the cell size limit 
> (0.900)
>
>
> 2 .Then I tried to increase the --dds and --rdd from original values 
> of 1.25 and 0.623 to 1.30 and 0.670 respectively.
>
>    But it does not help and ended with run crash.
>
> /*What I did was logical or I did it wrongly?*/
>
> /*Now can anyone please suggest me the appropriate way to deal with 
> the problem mentioned above? */
>
> As I want the continuation of the same run without altering the output 
> after change in the parameters (As I have to compare the output with 
> unbound protein run thus can't afford change in output with change in 
> parameters)
>
> I know that I need to change some of the parameters in .mdp file such as
>
> fourierspacing from 0.16 to 0.12 and on the contrary increase the 
> pme_order from say 4 to 6.
>
> /*But as asked above by doing so the output will not or will be the 
> exact continuation run?*/
>
> /*How to increase the ratio of cell size to constrain length per error 
> message?*/
>
> /*If any better way of doing so without changing the output please 
> suggest,*/
>
> I am suffering from the same problem since long,
>
> Please help me .Please see the mdp file for the reference.
>
> integrator        = md
>
> nsteps             = 10000000
>
>  dt                         = 0.002                   ; 2 fs
> > >
>  ; Output control
> nstxout                        = 1000                        ;
> nstvout                         = 1000                        ;
> nstxtcout                     = 1000                        ;
>
> nstenergy                    = 1000                        ;
>
>  nstlog              = 1000                        ;
> ; Bond parameters
> continuation   = yes                   ; Restarting after NPT
> constraint_algorithm = lincs ; holonomic constraints
> constraints     = all-bonds             ; all bonds (even heavy atom-H 
> bonds)
> lincs_iter         = 1                               ; accuracy of LINCS
> lincs_order     = 4
>
> ; Neighborsearching
> ns_type           = grid
> nstlist              = 5
>
>  rlist                = 1.2
> rcoulomb        = 1.2
> rvdw                = 1.2
> ; Electrostatics
> coulombtype  = PME
> pme_order     = 4
> fourierspacing           = 0.16
> tcoupl              = Nose-Hoover
> tc-grps                        = Protein  P    SOL_NA_CL ;
> tau_t                = 0.5   0.5       0.5
> ref_t                = 323 323     323  group, in K
> ; Pressure coupling is on
> pcoupl             = Parrinello-Rahman     ;
>
> pcoupltype     = semiisotropic
> tau_p              = 2.0                                   ; time
> ref_p               = 1.0   1.0                           ; reference
> compressibility = 4.5e-5       4.5e-5 ;
>
>  ; Periodic boundary conditions
> pbc                      = xyz                        ; 3-D PBC
>
>  ; Dispersion correction
> DispCorr        = EnerPres    ; account for cut-off vdW scheme
>
> ; Velocity generation
>   gen_vel                       = no
>
> Thanking in Advance
>
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20120426/7b577a01/attachment.html>