Quantum Reupload Units (QRU)

A hardware-efficient, single-qubit quantum architecture tailored for time series forecasting. This repository provides code, benchmarks, and analysis for the models described in the QCE25 paper: "Quantum Reupload Units: A Scalable and Expressive Approach for Time Series Learning".

📄 Paper

Quantum Reupload Units: A Scalable and Expressive Approach for Time Series Learning

✨ Highlights

• Single-qubit quantum architecture with re-uploaded inputs
• Demonstrates superior expressivity and convergence over PQC, VQC, and RNNs
• Benchmarked on both synthetic and real-world time series datasets
• Fourier-based expressivity and absorption witness analysis

🧠 Architectures

• QRU (Quantum Reupload Unit)
• PQC (Parameterized Quantum Circuit)
• VQC (Variational Quantum Circuit)
• QRU–QRB–Local (with shared ancilla)
• QRU–QRB–Global (with independent ancilla)

📊 Datasets

Dataset Source Mackey-Glass Synthetic chaotic time series Sinusoidal Wave Simple periodic series River Level TAIAO Project (Real-world dataset)

📌 Features

✅ Supports multiple quantum architectures: PQC, VQC, QRU, QRU-QRB (local/global)

✅ Fourier expressivity analysis via amplitude spectrum

✅ Absorption witness & KL divergence computation

✅ Realistic comparison with RNN using parameter-matched setup

✅ Easy extensibility for custom datasets

📖 Citation

@inproceedings{casse2025qru,
  title={Quantum Reupload Units: A Scalable and Expressive Approach for Time Series Learning},
  author={Cassé, Léa and Ponnambalam, Sabarikirishwaran and Pfahringer, Bernhard and Bifet, Albert},
  booktitle={IEEE Quantum Week (QCE25)},
  year={2025}
}

🧠 Authors

Léa Cassé – University of Waikato & École Polytechnique

Sabarikirishwaran Ponnambalam – Griffith University

Bernhard Pfahringer – University of Waikato

Albert Bifet – University of Waikato & Télécom Paris

🏷️ Tags & Topics

#QuantumML #TimeSeriesForecasting #Pennylane #QML #NISQ #QiskitCompatible #QuantumExpressivity #FourierAnalysis #GradientFlow #QuantumCircuitDesign #AbsorptionWitness

📜 License This project is licensed under the MIT License – see the LICENSE file for details.