[gmx-users] mdrun crashed with tables

[ms]

On 23/11/10 21:42, Sai Pooja wrote:
> Hi,
>
> I run a simulation of a protein in water(tip3p) system using charmm
> forcefield. I have specified the following energy groups:
> Protein SOL and energy group tables Protein SOL SOL SOL. I set Coulombtype
> to User and VdW to Cut-off. I then generate tables as specified in the
> manual. To test the validity of these tables, I standard Coulomb and LJ6,12
> tables.
> I generate 4 tables - table.xvg, table_Protein_SOL.xvg, table_SOL_SOL.xvg
> and tablep.xvg (all identical)
> When I run grompp I get no errors. When I run mdrun, it stops after the
> first step as LINCS fails to converge.
> I run the same simulation by changing the Coulombtype to Cut-off and without
> the tables and it runs.
>
> Can you suggest what could be going wrong?

Are the tables correct? The derivative is correct? Signs, units, etc.? 
How did you generate them?

Often the devil is in the most trivial -and therefore overlooked- errors.

What does the mdrun log say?

In any case, you can try using the -debug option of mdrun and have a 
look at the *interpolated* tables in output to see if they are 
consistent with your intentions (warning: using -debug can spit out 
gigabytes of stuff).

> (python script)
> Command used: cmd += '(%s/mdrun_mpi -s rex_%d.tpr -e rex_%d -c after_rex_%d
> -cpi restart%d -cpo restart%d -cpt 0.1 -append -g rexlog%d -x rextraj%d
> -table table.xvg table_Protein_SOL.xvg table_SOL_SOL.xvg -tablep tablep.xvg
>> /dev/null); ' %(GROMPATH,i,i,i,i,i,i,i)
>
> I tried this in two ways -
> 1) specified [ nonbonded_param ] for all Protein atoms and tip3p atoms
> 2) Not specifying them separately
>
> Thanks and regards
> Pooja
>


-- 
Massimo Sandal, Ph.D.
http://devicerandom.org