[gmx-users] LINCS error- 4.5.4 installation?

[Mark Abraham]

On 4/30/2011 9:06 AM, Juliette N. wrote:
> Dear experts,
>
> I just tried the command  mpirun -np 4 mdrun on 4.5.4 but after a few 
> steps I get LINCS error:
>
> step 400, will finish Sat Apr 30 05:20:45 2011
> Step 750, time 1.5 (ps)  LINCS WARNING
> relative constraint deviation after LINCS:
> rms 10.116560, max 72.359520 (between atoms 314 and 315)
> bonds that rotated more than 30 degrees:
>  atom 1 atom 2  angle  previous, current, constraint length
>     301    302   61.0    0.1090   0.1841      0.1090
>
>
> I tried to run using the same input files on 4.0.7 and simulation 
> worked fine. I mean the LINCS is not because of flawed topology or 
> instability issue. I am wondering this is happening because of MPI or 
> installation ..Please help!

Normally, a successful MPI installation would have produced an mdrun_mpi 
executable, not mdrun. 15 minutes before this email you didn't manage to 
configure successfully with --enable-mpi. So probably you are not 
running in parallel. Look at the number of .log files produced, and 
their first line.

Regarding the LINCS error, your simulation was probably marginally 
stable, and got lucky with the integration in 4.0.7 and unlucky in 4.5. 
That's life - you should consider the management strategies here 
http://www.gromacs.org/Documentation/Terminology/Blowing_Up

Mark