SDE discovery code

Author: arshednabeel

.DS_Store

@@ -0,0 +1,2 @@
+.DS_Store
+__pycache__

.gitignore

@@ -1,5 +1,15 @@
 # Data driven mesoscale SDE
 
+This repository contains the code for the manuscript:
+
+Data-driven discovery of stochastic dynamical equations of collective motion. 
+Nabeel, A., Jadhav, V., Sire, C., Theraulaz, G., Escobedo, R., Iyer, S. K., & Guttal, V. (2023).
+DOI: 10.1088/1478-3975/ace22d
+
+The code is divided into two parts:
+  - The directory `/spp_model` contains MATLAB code to simulate the agent-based models.
+  - The directory `/sde_discovery` contains code to discover mesoscale SDEs from the simulated trajectories.
+
 ## Simulating the flocking model
 
 **Generate data of positions and velocities to calculate group polarisation for given group size and parameters.**
@@ -12,6 +22,10 @@ To simulate stochastic pairwise interaction set `k_alg = 1` in `sim_data.m`. Sim
 
 After running the simulation, use the data in `n_pw.mat` to calculate the order parameter, i.e., group polarisation. To do so, run the Matlab code `/figures/grp_pol.m`. This file stores all the required data in the `n_pw.csv` file in the format `[mx, my, m]`.
 
-## Simulation video
+### Simulation video
 
 To see the collective motion of agents, run the code `/spp_model/simulations.m`. Make sure to load `n_pw.mat`. Variables are defined within the code and can be changed accordingly.
+
+## Data driven SDE discovery
+
+The notebook `/sde_discovery/sde-discovery.ipynb` contains code to analyze the order-parameter time series and discover mesoscale stochastic differential equations. The code uses the CSV file generated from the MATLAB code above. An example CSV file, corresponding to the _ternary local interaction model_ is provided.