# Multiple Walker simulations without MPI

**URL:** <https://gromacs.bioexcel.eu/t/multiple-walker-simulations-without-mpi/3683>\
**Category:** User discussions\
**Tags:** mdrun\
**Created:** [February 17, 2022, 3:33am UTC](https://gromacs.bioexcel.eu/t/multiple-walker-simulations-without-mpi/3683 "2022-02-17T03:33:33Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![plagueis](https://avatars.discourse-cdn.com/v4/letter/p/3be4f8/32.png) [@plagueis](https://gromacs.bioexcel.eu/u/plagueis)\
**Post date:** [February 17, 2022, 3:33am UTC](https://gromacs.bioexcel.eu/t/multiple-walker-simulations-without-mpi/3683/1 "2022-02-17T03:33:33Z")

</div>

GROMACS version: 2019.4  
GROMACS modification: Yes, PLUMED

Hi all,

The PLUMED Belfast tutorial for replica-exchange and multiple walker simulations ([PLUMED: Belfast tutorial: Replica exchange II and Multiple walkers](https://www.plumed.org/doc-v2.6/user-doc/html/belfast-8.html)) notes that multiple walker simulations can be run independently without MPI, using the filesystem to share the biasing potential. Does this require any different directory structure to the -multi method? And how would the mdrun call look like?

My guess is from the original:

> mpirun -np 4 gmx\_mpi mdrun -s topol -plumed plumed -multi 4 \>& log &

Instead:

> gmx mdrun -s topol0.tpr -plumed plumed.dat  
> gmx mdrun -s topol1.tpr -plumed plumed.dat  
> gmx mdrun -s topol2.tpr -plumed plumed.dat  
> gmx mdrun -s topol3.tpr -plumed plumed.dat

Is this correct?