Create a pool of MAPDL instances#

PyMAPDL contains the LocalMapdlPool class to simplify creating multiple local instances of the Mapdl class for batch processing. This can be used for the batch processing of a set of input files, convergence analysis, or other batch related processes.

This code creates a pool:

>>> from ansys.mapdl.core import LocalMapdlPool
>>> pool = LocalMapdlPool(10)
'MAPDL Pool with 10 active instances'
>>> pool.exit(block=True)

You can supply additional keyword arguments when creating the pool. This code creates several instances with one CPU each running at the current directory within their own isolated directories:

>>> import os
>>> my_path = os.getcmd()
>>> pool = LocalMapdlPool(10, nproc=1, run_location=my_path)
Creating Pool: 100%|########| 10/10 [00:01<00:00,  1.43it/s]

You can access each individual MAPDL instance with this code:

>>> pool[0]
<ansys.mapdl.core.mapdl.MapdlGrpc object at 0x7f66270cc8d0>

Note that this is a self-healing pool. If an instance of MAPDL dies during a batch process, that instance is automatically restarted. You can turn off this behavior by setting restart_failed=False when creating the pool.

>>> from ansys.mapdl.core import examples
>>> files = [examples.vmfiles["vm%d" % i] for i in range(1, 21)]
>>> outputs = pool.run_batch(files)
>>> len(outputs)
20

completed_indices = []


def func(mapdl, input_file, index):
    # input_file, index = args
    mapdl.clear()
    output = mapdl.input(input_file)
    completed_indices.append(index)
    return mapdl.parameters.routine


inputs = [(examples.vmfiles["vm%d" % i], i) for i in range(1, 10)]
output = pool.map(func, inputs, progress_bar=True, wait=True)
[
    "Begin level",
    "Begin level",
    "Begin level",
    "Begin level",
    "Begin level",
    "Begin level",
    "Begin level",
    "Begin level",
    "Begin level",
]

# Close the PyMAPDL pool.
pool.exit()

>>> pool.exit()