About

This repository holds the preprint and code for Kern, Date, Chao, 2024: Effects of Spatial Constraints of Inhibitory Connectivity on the Dynamical Development of Criticality in Spiking Networks.

To recreate the figures

wihtout running any simulations, or to play around with our data, follow these steps:

1. Clone this repository

2. Set up a python environment with the dependencies listed in requirements.txt, e.g.:

    python -m venv .env
    pip install -r requirements.txt
   

3. Download the abridged data from https://doi.org/10.6084/m9.figshare.26796061 and unzip it as a results/ subdirectory

4. Run all jupyter notebooks to generate the main and supplementary figures.

To replicate the published results without data

To run new simulations, I recommend using a Linux or similar system. Windows users are recommended to use WSL. Simulation and analysis are possible in any system, but will require a little more footwork to replicate the work the bash scripts (steps 3-4 below) do. Luckily, the scripts are very simple (go look), and are easily replaced with some copy-paste drudgery. If that's not for you, follow these steps:

1. Clone this repository

2. Set up a python environment with the dependencies listed in requirements.txt, e.g.:

    python -m venv .env
    pip install -r requirements.txt
   

3. Run simulations with a specified runtime seed, which should be a positive integer:

    bash grid-run.sh <runseed>
   

   Note:

   1. This runs through all combinations of p_inh and r_inh as defined in the parameter files in params/. Note that each parameter file itself defines 10 networks to be run in parallel (N_nets=10), generated with a consistent random seed for structure (rng=0).
   2. To recreate the published results exactly (minus possible floating point inaccuracies etc.), call grid_run.sh (and subsequent scripts) with runseeds 0, 100, 200, 300, and 400 in five separate calls.
   3. Every run of a single parameter setting generates ~1.3G of raw data; i.e., invoking grid-run.sh once requires on the order of ~35G of disk space.
   4. Caution: There is no overwrite protection. If you are not careful with the naming scheme, the contents of existing subdirectories in results/ may be overwritten.

4. Run analysis scripts to munge the raw data into manageable size, again using the runseed as an integer argument:

    bash grid-spikestats.sh <runseed>
    bash grid-wstats.sh <runseed>
   

   Notes (i) and (ii) above apply here, too.

5. Run postprocessing script for further data munging:

    python grid_postprocessing.py <runseed[s]>
   

   Unlike the bash scripts, this can be called with multiple runseed arguments (e.g., python grid_postprocessing.py 0 100 200 300 400), but note that it relies on the outputs of step 4 above.

6. Run all jupyter notebooks to generate the main and supplementary figures.