4.7: Parallelism: Using the Joblib backend

This example demonstrates how to use the Joblib backend for simulating dipoles using hnn_core.

hnn_core can take advantage of the Joblib library to run multiple independent simulations simultaneously across multiple CPU processors. This is an example of "embarrassingly parallel" processing jobs. In HNN, this is commonly done if you want to run many "trials" of the same simulation. Since each trial simulation is fully independent of the other trial simulations, each trial simulation can be run on its own CPU core. Joblib parallelism is particularly useful if you are using batch simulation to explore parameter spaces.

Note that to use Joblib parallelism, you need either the conda install or the Joblib Installation dependencies described in our Installation Guide here.

Note that Joblib parallelism is distinct from hnn_core's use of MPI parallelism, which can be found here.


        # Authors: Mainak Jas 
#          Blake Caldwell 
#          Austin Soplata

Let us import what we need from hnn_core:


        import matplotlib.pyplot as plt

from hnn_core import simulate_dipole, jones_2009_model
from hnn_core.viz import plot_dipole

Following our Alpha example, we will create our network and add a ~10 Hz "bursty" drive:


        net = jones_2009_model()

weights_ampa = {'L2_pyramidal': 5.4e-5, 'L5_pyramidal': 5.4e-5}
net.add_bursty_drive(
    'bursty', tstart=50., burst_rate=10, burst_std=20., numspikes=2,
    spike_isi=10, n_drive_cells=10, location='distal',
    weights_ampa=weights_ampa, event_seed=278)

Finally, we will simulate using the JoblibBackend class. You can control the number of CPU cores to use via n_jobs, while the number of total trials to be run can be specified by n_trials. Note that these numbers do NOT have to match: you can ask for more trials than there are jobs available, and Joblib will simply execute later jobs after the first batch has completed.


        from hnn_core import JoblibBackend

with JoblibBackend(n_jobs=4):
    dpls = simulate_dipole(net, tstop=210., n_trials=6)

Out:

Joblib will run 6 trial(s) in parallel by distributing trials over 4 jobs. Building the NEURON model Building the NEURON model Loading custom mechanism files from /opt/anaconda3/envs/website-redesign-mpi/lib/python3.12/site-packages/hnn_core/mod/arm64/.libs/libnrnmech.so Building the NEURON model Loading custom mechanism files from /opt/anaconda3/envs/website-redesign-mpi/lib/python3.12/site-packages/hnn_core/mod/arm64/.libs/libnrnmech.so Building the NEURON model [Done] Trial 2: 0.03 ms... [Done] Trial 4: 0.03 ms... [Done] [Done] Trial 3: 0.03 ms... Trial 1: 0.03 ms... Trial 2: 10.0 ms... Trial 4: 10.0 ms... Trial 1: 10.0 ms... Trial 3: 10.0 ms... Trial 2: 20.0 ms... Trial 4: 20.0 ms... Trial 3: 20.0 ms... Trial 1: 20.0 ms... Trial 2: 30.0 ms... Trial 4: 30.0 ms... Trial 3: 30.0 ms... Trial 1: 30.0 ms... Trial 2: 40.0 ms... Trial 4: 40.0 ms... Trial 3: 40.0 ms... Trial 1: 40.0 ms... Trial 2: 50.0 ms... Trial 4: 50.0 ms... Trial 3: 50.0 ms... Trial 1: 50.0 ms... Trial 2: 60.0 ms... Trial 4: 60.0 ms... Trial 3: 60.0 ms... Trial 1: 60.0 ms... Trial 2: 70.0 ms... Trial 4: 70.0 ms... Trial 3: 70.0 ms... Trial 1: 70.0 ms... Trial 2: 80.0 ms... Trial 4: 80.0 ms... Trial 3: 80.0 ms... Trial 1: 80.0 ms... Trial 2: 90.0 ms... Trial 4: 90.0 ms... Trial 3: 90.0 ms... Trial 1: 90.0 ms... Trial 2: 100.0 ms... Trial 4: 100.0 ms... Trial 3: 100.0 ms... Trial 1: 100.0 ms... Trial 2: 110.0 ms... Trial 4: 110.0 ms... Trial 3: 110.0 ms... Trial 1: 110.0 ms... Trial 2: 120.0 ms... Trial 4: 120.0 ms... Trial 3: 120.0 ms... Trial 1: 120.0 ms... Trial 2: 130.0 ms... Trial 4: 130.0 ms... Trial 3: 130.0 ms... Trial 1: 130.0 ms... Trial 4: 140.0 ms... Trial 2: 140.0 ms... Trial 3: 140.0 ms... Trial 1: 140.0 ms... Trial 4: 150.0 ms... Trial 2: 150.0 ms... Trial 3: 150.0 ms... Trial 1: 150.0 ms... Trial 2: 160.0 ms... Trial 4: 160.0 ms... Trial 3: 160.0 ms... Trial 1: 160.0 ms... Trial 4: 170.0 ms... Trial 2: 170.0 ms... Trial 3: 170.0 ms... Trial 1: 170.0 ms... Trial 4: 180.0 ms... Trial 2: 180.0 ms... Trial 3: 180.0 ms... Trial 1: 180.0 ms... Trial 4: 190.0 ms... Trial 2: 190.0 ms... Trial 3: 190.0 ms... Trial 1: 190.0 ms... Trial 4: 200.0 ms... Trial 2: 200.0 ms... Trial 3: 200.0 ms... Trial 1: 200.0 ms... Building the NEURON model Building the NEURON model [Done] [Done] Trial 6: 0.03 ms... Trial 5: 0.03 ms... Trial 6: 10.0 ms... Trial 5: 10.0 ms... Trial 6: 20.0 ms... Trial 5: 20.0 ms... Trial 6: 30.0 ms... Trial 5: 30.0 ms... Trial 6: 40.0 ms... Trial 5: 40.0 ms... Trial 6: 50.0 ms... Trial 5: 50.0 ms... Trial 6: 60.0 ms... Trial 5: 60.0 ms... Trial 6: 70.0 ms... Trial 5: 70.0 ms... Trial 6: 80.0 ms... Trial 5: 80.0 ms... Trial 6: 90.0 ms... Trial 5: 90.0 ms... Trial 6: 100.0 ms... Trial 5: 100.0 ms... Trial 6: 110.0 ms... Trial 5: 110.0 ms... Trial 6: 120.0 ms... Trial 5: 120.0 ms... Trial 6: 130.0 ms... Trial 5: 130.0 ms... Trial 6: 140.0 ms... Trial 5: 140.0 ms... Trial 6: 150.0 ms... Trial 5: 150.0 ms... Trial 6: 160.0 ms... Trial 5: 160.0 ms... Trial 6: 170.0 ms... Trial 5: 170.0 ms... Trial 6: 180.0 ms... Trial 5: 180.0 ms... Trial 6: 190.0 ms... Trial 5: 190.0 ms... Trial 6: 200.0 ms... Trial 5: 200.0 ms...


        plot_dipole(dpls, show=False)
plt.show()

Out: