[gmx-users] job magic number in MXMPI_MAGIC

[Mark Abraham]

Sheyore Omovie wrote:
> Thanks Tom,
> I have tried that in the past though, and I still get Epot values of nan.

One solution is to take your (successful) individual .top files and edit 
them in line with the example in chapter 5 to create .itp files for your 
two molecules. Then include them both in a master .top, set up a 
structure file with both molecules and use editconf and genbox and 
grompp in the usual way.

> It seems the only other option I have is to place the molecules in 
> separate boxes, set
> disre=simple, and use the -multi option during the mdrun. The -multi 
> option only works with the parallel installation.

-multi allows you to do multiple independent simulations. This won't do 
anything differently for you than doing the independent simulations 
asynchronously.

> Unfortunately I get the ff error message when I try to use the parallel 
> gromacs installation on the server:
> /<MPICH-MX> Error: Need to obtain the job magic number in MXMPI_MAGIC./
> /I would appreciate any advice on how to fix this./

This is an MPI problem, not a GROMACS problem. Consult your sysadmins.

Mark