On 11/5/17 7:13 AM, מירב בריטברד wrote: > I managed to build with MPI but got a runtime error (below)

This sounds like the usual OpenMPI problem documented in the IMP manual (with workarounds): https://integrativemodeling.org/2.8.0/doc/ref/namespaceIMP_1_1mpi.html

> is it critical to run replica exchange in order to get the best results > or Monte Carlo is also ok?

If you don't have MPI you'll just get replica exchange with one replica, which will give you less complete sampling, of course.

Ben