grid_disruption

I. Experimental Data Analysis

Intended directory structure:

  ├── Code-for-Ying-et-al.-2023
  |    ├── extracted_all: experimental data converted into python-readable format.
  |    ├── Figure_1
  |    ├── Figure_2
  |    └── ...
  └── grid_disruption: this directory

Data Preparation

0. Make sure MATLAB is installed (preferably 2021b).
1. Clone Code-for-Ying-et-al.-2023 repository and place data_converter_all.m inside the Figure_2 directory.
2. Clone CMBHOME directly under the main directory (Code-for-Ying-et-al.-2023).
3. Launch MATLAB and run the following commands from inside Figure_2 to convert the data to python-readable format.

addpath ../session_data
addpath ../session_data2
addpath ../CMBHOME
import CMBHOME.*
data_converter_all

Data Processing

Run ying2023_ratemap.py to calculate the ratemaps as well as different metrics (example shell script: ying2023_ratemap.sh).

The results are stored inside the data directory.

Figure Production

analysis_method.ipynb: Figures to help explain the method analysis_heatmap.ipynb: Heatmap analysis_ecdf.ipynb: Empirical Cumulative Distributions

CNN analysis

Run classifier.py to train a convolutional neural network.

analysis_cnn.ipynb produces corresponding figures.

II. RNN Perturbation Analysis

1. Train the RNN by running main.py (example shell script: run.sh).
2. Run eval.py to perform perturbations (example shell script: eval.sh).
3. analysis_rnn.ipynb produces corresponding figures.