Circuit Fingerprint Prediction (MIT iQuHack 2026)

Team Submission for Quantum Rings Challenge Automated prediction of quantum circuit runtime and optimization thresholds using Random Forest ensembles.

🏆 Project Overview

This project was built for MIT iQuHack 2026 to solve the "Circuit Fingerprint" challenge. The goal is to predict:

1. Optimization Threshold: The ideal compression level (1-256) for a quantum circuit to achieve >0.75 fidelity.
2. Runtime: The wall-clock execution time of the circuit simulation.

Our solution uses a Chained Random Forest approach, where the predicted threshold is fed as a feature into the runtime predictor, achieving 66.7% accuracy on threshold classification and ~60s MAE on runtime prediction.

🚀 Key Features

• Interaction Distance: A custom feature metric measuring the topological complexity of qubit interactions.
• Chained Prediction: Pipeline that first determines the complexity class (Threshold) before estimating cost (Runtime).
• Hardware Aware: Feature sets include processor type (CPU/GPU) and precision (Single/Double).
• Robustness: Handled missing data and verified on unseen "Holdout" circuits with 100% pipeline stability.

🛠️ Usage

Quick Start

1. Install Dependencies:

   pip install -r requirements.txt

2. Run Prediction:

   python predict.py --tasks path/to/tasks.json --circuits path/to/qasm/dir --out predictions.json

Repository Structure

• predict.py: Main inference script.
• feature_ext.py: QASM feature extraction logic.
• models/: Pre-trained LightGBM/RandomForest models.
• train_model.py: Training pipeline (with strict validation split).
• writeup.md: Detailed technical report.

📊 Performance

Metric	Score	Note
Threshold Accuracy	66.67%	Exact class match
Runtime MAE	62.80s	Mean Absolute Error

👥 Contributors

• HABER7789 (and Team)