readme

introduction

This program simulates a spiking neural network of 12500 neurons described in Nicolas Brunel's paper "Dynamics of Sparsely Connected Networks of Excitatory and Inhibitory Spiking Neurons".

The main objective is to be able to reproduce figure 8 found on page of the abovementioned paper. To do so, I've implemented a C++ program that outputs a data file containing the spiking times of 50 randomly chosen neurons.

implementation documentation

All information concerning the C++ program can be found in the attributed Doxygen file. cppcourse-brunel/doc/html/index.html Additional information about the simulation can be found in the paper. Doxygen also incorporated into Cmake.

running the program

The program is automatically set to case C of figure 8. In order to reproduce all cases, constants G and ETA are to be changed in the constants.h file. Running the simulation will print out the number of average spikes per second.

Note: while the 1 second simulation on my personal computer has a total runtime averaging 40 seconds, when tested on EPFL's machines the runtime can often double (usually around 2 minutes - depending on the G and ETA)

How to run :

Clone git repository
Go to cppcourse-brunel file
Open terminal
type cd build in order to enter build folder (if folder does not exist, type in mkdir build beforehand)
type cmake ../ as CMakeList is in main folder
type make
run target ./brunel or ./bruneltest for gtest
Data file will be in results folder
Plotting can be done with jupyter or website (c.f. below)
To clean, type in rm -r *

plot reproduction

In order to plot the data, I developed a jupyter notebook file creating two plots : a scatterplot and a histogram.

The scatterplot shows the neuron ID (y-axis) depending on the spiking time (x-axis). In fact, every dot corresponds to one spike.
The histogram shows the number of spikes for a given time interval (bins of 1 ms).

The results in this file are from my own jupyter notebook, but equivalent results could be obtained from either Prof. Gewaltig's jupyter notebook or Antoine Albertelli's web application https://cs116-plot.antoinealb.net/.

jupyter notebook

here is the specific jupyter notebook I used. The ipynb file can be found in cppcourse-brunel/results/rasterplot.ipynb and in cppcourse-brunel/build/rasterplot.ipynb. I suggest running the notebook from the build file as it searches for the data in it's directory, and the data is saved in build directory.

rasterplot.ipynb

import matplotlib.pyplot as plt
import numpy as np

x, y = np.loadtxt('rasterdata.gdf', delimiter = '\t', unpack = True)

fig,ax = plt.subplots(1)
ax.plot(x,y, "|", alpha=.8)
plt.xlim([500, 600])
plt.ylabel('Neuron ID')
ax.set_xticklabels([])
plt.savefig('fig.png', dpi = 400)
plt.show()

plt.hist(x, 1000, rwidth=1, alpha=.75)
plt.xlim([500, 600])
plt.ylim([0,50])
plt.xlabel('Time in ms')
plt.ylabel('Spike amount')
plt.grid(True)
plt.savefig('hist.png', dpi = 400)
plt.show()