Description

Experiment and visualization code for the paper: "Parametrizing Analog Multi-Compartment Neurons with Genetic Algorithms" (in review).

The main part of the experiment is to find the appropriate inter-compartment conductance and leak conductance on the neuromorphic computing platform BrainScaleS-2 using genetic algorithms to replicate a desired attenuation behavior of an excitatory post synaptic potential propagating along a linear chain of compartments.

You can access BrainScaleS-2 via EBRAINS. There you can also find a simple tutorial on how to use genetic algorithms to parameterize BrainScaleS-2.

Functionality

All experiments of the paper are summerized in one script and can be executed via (assuming you have access to BrainScaleS-2):

python3 src/py/model_paper_mc_genetic/scripts/record_experiment.py --grid_search --genetic_algorithm

Using the recorded data you can visualize it by executing:

python3 src/py/model_paper_mc_genetic/scripts/plot_experiment.py --latex

If you don't have access to BrainScaleS-2 you can also download the data of the publication from heiDATA and replicate the figures of the paper using the previous command.

Structure

In src/py/model_paper_mc_genetic/

• scripts/record_experiment.py executes all experiments of the paper.
• scripts/plot_experiment.py visualizes the data recorded in record_experiment.py
• scripts/attenuation_ga.py provides the genetic algorithm applied to the attenuation experiment.
• plotting.py helper functions for plotting the data.
• helpers.py further helper functions.