Toward Large-Scale Distributed Quantum Long Short-Term Memory with Modular Quantum Computers

Published in IWCMC 2025 – The 21st International Wireless Communications & Mobile Computing Conference, 2025

This paper presents a Distributed Quantum Long Short-Term Memory (QLSTM) framework designed for execution on modular quantum computing architectures, enabling scalability on Noisy Intermediate-Scale Quantum (NISQ) devices. Highlights include:

Embedding variational quantum circuits (VQCs) into LSTM cells to capture long-range temporal dependencies.
Partitioning VQCs into subcircuits for execution across a network of Quantum Processing Units (QPUs).
Validation on challenging sequence tasks, such as damped harmonic oscillators and NARMA sequences, demonstrating stable convergence and enhanced training efficiency.

This approach lays the groundwork for integrating distributed quantum learning into future hybrid quantum-classical HPC systems, supporting large-scale AI applications in dynamic environments.

Figure: Modular QPU architecture and benchmark results for Distributed QLSTM applied to temporal sequence modeling.

Recommended citation: Chen, Kuan-Cheng; Chen, Samuel Yen-Chi; Liu, Chen-Yu; & Leung, Kin K. (2025). "Toward Large-Scale Distributed Quantum Long Short-Term Memory with Modular Quantum Computers." Proceedings of IWCMC 2025 – The 21st International Wireless Communications & Mobile Computing Conference.
Download Paper