🏆 QPU Circuit Packing – Winning Submission at Q-SITE Hackathon 2024

🚀 Overview

This repository contains the first-place winning submission for the QPU Circuit Packing Challenge at the Q-SITE Hackathon 2024. Our project introduces a hardware-aware approach to packing multiple copies of quantum circuits onto a single QPU execution, maximizing efficiency while maintaining high fidelity.

---

🧠 Problem Statement

Most quantum algorithms underutilize today's large QPUs. For example, running a 20-qubit circuit on IBM’s 127-qubit Eagle chip wastes over 100 idle qubits.

Goal:
Develop a strategy to replicate and pack a user's small circuit multiple times on a QPU, reduce the required number of shots, and control fidelity degradation.

---

🎯 Key Objectives

• 🧩 Replicate the input quantum circuit n times.
• ⚙️ Transpile each copy to disjoint, high-fidelity subregions of the QPU.
• 🔬 Execute all copies in one job, reducing total shots.
• 📊 Recover and separate measurement results with minimal fidelity loss.

---

🔍 Technical Highlights

✅ Qubit Placement

• Uses IBM's coupling map and Layout2qDistance to assign the best zones to each circuit.
• Avoids "noisy" or poorly connected qubits, guided by modularity-based community detection.

🔁 Packing Strategy

• Greedy + Graph Partitioning hybrid algorithm to identify optimal qubit clusters.
• Adaptive to backend architecture (e.g., IBM Fez, Eagle).

📉 Crosstalk Mitigation

• Spatial separation enforced during packing to reduce coherent noise from RF interference.

---

📈 Result

Our method demonstrates:

• 📦 Efficient QPU utilization (up to 9x packing on 127 qubits).
• 🎯 Fidelity loss < 5% across multiple circuit copies.
• ⏱️ 5x–9x reduction in total runtime.

---

📂 Contents

• notebook.ipynb: Main implementation and experiment notebook.
• packing_algorithm.py: Core logic for zone partitioning and layout.
• results/: Benchmark results, fidelity plots, and comparisons.
• ibm_fez.jpg: Reference chart for hardware-aware error trends.

---

🧪 Tools & Technologies

• 🧱 Qiskit for circuit definition and transpilation.
• 🔍 networkx for backend graph analysis.
• 📦 IBM Q backend APIs (via qiskit-ibm-runtime) for live deployment.
• 📊 Matplotlib for visualization and fidelity analysis.

---

📜 Citation

QPU Circuit Packing: A Fidelity-Aware Multi-Copy Execution Strategy.
🏆 Winner – Q-SITE Hackathon 2024
Authors: Hemish Ahuja, Mukul, Maral, Negar

---

📅 Last Evaluated:

August 31, 2024 — on IBM Fez

---

🤝 Acknowledgments

Huge thanks to Haiqu, IBM Quantum, and the Q-SITE Hackathon organizers for hosting this cutting-edge quantum challenge.

---

📬 Contact

Feel free to reach out or connect on LinkedIn if you want to collaborate or learn more!