matensemble 0.5.3

An adaptive and highly asynchronous ensemble simulation workflow manager MatEnsemble (https://github.com/Q-CAD/MatEnsemble) built jointly on top of the hierarchical graph based scheduler FLUX and concurrent-futures infrastructure of python

MatEnsemble

MatEnsemble is a Python library for high-throughput workflows on HPC systems. You define a directed acyclic graph (DAG) of chores—Python callables or executable commands—and MatEnsemble submits work through Flux, tracks completions, adapts scheduling to free CPUs and GPUs, and writes structured logs and per-chore output directories.

Here is an example which defines a simple MPI chore, adds ten independent instances of it to a pipeline, and submits the workflow.

pipe = Pipeline()

# register a function the MatEnsemble
@pipe.chore(num_tasks=10, cores_per_task=1, gpus_per_task=0, mpi=True)
def mpi_hello_world():
    size = MPI.COMM_WORLD.Get_size()
    rank = MPI.COMM_WORLD.Get_rank()
    name = MPI.Get_processor_name()

    print(f"Hello World! I am process {rank} of {size} on {name}.")

# adding 10 mpi_hello_world chores to the pipeline
for _ in range(10):
    mpi_hello_world()

pipe.submit(log_delay=1)

Rather than launching the MPI jobs directly, you describe the work declaratively with the @pipe.chore decorator and enqueue each invocation. MatEnsemble then schedules the chores through Flux, allocating the requested resources (num_tasks, CPU cores, GPUs, and MPI support) for each job.

Installation

If you are trying to use MatEnsemble on the Frontier, Pathfinder or Perlmutter super computers then you can quickly install MatEnsemble with our install script:

curl -fsSL https://raw.githubusercontent.com/FredDude2004/MatEnsemble/refs/heads/main/install.sh | bash

For more general installation see our documentation

Related

• Flux documentation
• Flux Python guide
• Slurm documentation (common front-end to batch allocations)
• LAMMPS manual (often used alongside ensemble MD workflows)

License

BSD 3-Clause. See LICENSE.