On Tue, Aug 14, 2018, emanuele wrote:
> I am trying to get the free energy for a reduction A + e- → A- using the
> Thermodynamic Integration (TI) in SANDER.
>
> When running a classical TI in SANDER everything goes well, but I am
> unable to do the same when the system is treated at QM/MM level ( ORCA
> is used for the QM part).
>
> I am running sander in serial on two jobs :
>
> mpirun -n 2 /$AMBERHOME$/sander.MPI -ng 2 -groupfile model_step1.group "
>
> but the error, which I get, makes me think that there's a problem with
> recalling MPI .
>
> "Open MPI does not support recursive calls of mpirun" (ORCA error).
>
> In fact, I know that multisander works calling MPI, so when using more
> than 1 CPU-core in ORCA (Quantum package), there is a conflict in
> calling recursively MPI.
Exactly: both sander and ORCA are trying to run in parallel, using the
same MPI package.
>
> Do you have any idea about fixing the problem, ?
A workaround: since sander TI has to be run under MPI, use the serial
version of ORCA.
A possible(?) fix: install sander.MPI with mpich. ORCA is hard-wired to
use openmpi, so it's possible that this combination will work.
(Completely untested!!)
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Aug 14 2018 - 08:00:04 PDT
