Neural Resilient Pinning Control for Kuramoto Networks

---

🌐 Overview

This repository implements a Neural Resilient Pinning Controller (NRPC) for resiliente synchronization in Kuramoto oscillator networks under structural and dynamical perturbations. The goal is to learn which nodes to control in order to achieve a better resilience–energy trade-off, compared to classical full-actuation and heuristic strategies.

---

✨ Key Features

• 🎯 Learned node selection via neural networks

• ⚡ Energy-efficient control (sparse actuation)

• 🛡️ Resilience to structural disruptions

• 📊 Full evaluation pipeline

• 🔁 Reproducible experiments

---

📁 Repository Structure

NRPC-kuramoto/
├── kuramoto/
├── results/
├── figures/
├── stats/
├── train.py
├── eval.py
├── eval_fail.py
├── eval_hist_mask.py
├── dynamics_without_controller.py
├── requirements.txt
└── README.md

---

⚙️ Installation

python -m venv venv
source venv/bin/activate  # or venv\Scripts\activate on Windows
python -m pip install -r requirements.txt

---

🚀 Usage

Train:

python train.py

Evaluate:

python eval.py

Additional evaluation scenarios:

python eval_fail.py
python eval_hist_mask.py

---

📊 Outputs

• .pt trained models
• .pkl evaluation results
• .pdf figures

---

📈 Generate figures and statistics

Generates heatmaps, Pareto fronts, and statistical summaries.

python figures/pareto_eval.py
python figures/heatmap_ring.py
python figures/heatmap_diagonal_a.py
python figures/heatmap_diagonal_b.py
python stats/stats.py
python stats/stats_fail.py

---

🔁 Reproducibility

All results can be reproduced using the provided scripts.

---

📜 License

See LICENSE file.

Note

This repository provides the additional and reproducible material for the paper:

"Resilient Synchronization of Kuramoto Networks via Sparse Energy-Efficient Control",

by Manuela Chacon-Chamorro, María José Charria-Macías,
Luis Felipe Giraldo (Member, IEEE), and
Nicanor Quijano (Senior Member, IEEE),

submitted to the IEEE Conference on Decision and Control (CDC), 2026.

---