[gmx-users] Re: Excessive compute time using PME

[Jason de Joannis]

Try it in a cubic box. If it still explodes, you will know that it is
an unstable configuration with divergent energies. If it runs, then
you can start worrying about the box issue. There is no mdp parameter
for box-types.

/Jason de Joannis
Emory Univ, Atlanta

> Date: Thu, 19 May 2005 13:13:41 +0200
> From: "Jordi Camps" <jcamps at lsi.upc.edu>
> Subject: RE: [gmx-users] Excessive compute time using PME
> To: "'Discussion list for GROMACS users'" <gmx-users at gromacs.org>
> Message-ID: <001101c55c63$d148bd10$0200a8c0 at lan>
> Content-Type: text/plain; charset="us-ascii"
>
> Hello,
>
> I have some news about the problem. Yesterday I saw that the input that I
> was feeding was not completely contained in the octahedron, so I recentered
> the travelling waters and now I have a true octahedron as input for mdrun.
> If I chech the coordinates of the tpr file (extracted with editconf) I
> continue seeing an octahedron in VMD.
> But when I run the simulation with mdrun, it fails. And before core dumping,
> mdrun writes two files: step-1.pdb and step0.pdb.
> Step0.pdb shows an exploded molecule (and waters).
> Step-1.pdb shows my input in cubic boundaries. I think that this is the
> problem. I'm feeding an octahedron and mdrun is interpreting this like a
> box. There are some special considerations I should take into account if I
> want to simulate with an octahedron? Perhaps I'm missing some parameter in
> the grompp.mdp file? The box vectors, like the ones that I have in the file,
> are 6.72730 6.34256 5.49282 0.00000 0.00000 2.24243 0.00000 -2.24243
> 3.17128.
>
> Thank you for yout help.
>
> Sincerely,
>
> --
> Jordi Camps Puchades