Attractor dynamics for networks with learning rules inferred from in vivo data (U. Pereira & N. Brunel, Neuron 2018)

This repository contains three directories:

1. Dynamics
2. CapacityFung
3. MaximalCapacity

With a few modifications, this code reproduces most of the figures on the paper. Send me an e-mail if you have further inquiries.

Dynamics

It contains the code for reproducing Fig 6 A-F. The results on Fig 3 and Fig 4 A are reproduced by changing the value of the parameter A from 9.5 to 3.5 and setting the input current during the delay period as:

1. Standard normal iid for a novel stimulus
2. One of the stored patterns for familiar stimulus.

For generating fig4.pdf, first run python main.py on the numeric/network_simulation directory. The simulation will take a couple of hours depending on your machine. It will create two large files the_dynamics.p' (~800MB) and the_overlaps.p' (~1.15GB). They store the dynamics of 100 neurons and all the overlaps for 10 realizations of the 8s dynamics.

Capacity_vs_g

It contains the code for reproducing Fig 4 B-C. The same code can be used to perform the parameter exploration in Fig 5.

MaximalCapacity

It contains the code for reproducing S7.