Module-conditioned distribution of quantum circuits

2025-01-21

AI Summary

Distributed quantum computing (DQC) suffers from high inter-module communication overhead and lacks verifiable benchmarks for optimizing multi-module quantum circuit partitioning. Method: We propose the first exact integer linear programming (ILP) model for quantum circuit distribution under fixed module assignment, enabling co-optimization with arbitrary module allocation algorithms. Quantum circuits are modeled as hypergraphs, and inter-module gate communication costs are formally encoded; comparative analysis exposes inherent limitations of conventional hypergraph partitioning in DQC. Results: Experiments show that simple heuristic assignments achieve ILP-optimal solutions for representative circuits (e.g., QFT), and our approach significantly reduces communication overhead compared to state-of-the-art methods. This work establishes the first verifiable, reproducible post-compilation optimization framework for distributed quantum compilation.

Technology Category

• Constraint Satisfaction and Optimization: Distributed CSP/Optimization
• Machine Learning: Quantum Machine Learning
• Search and Optimization: Distributed Search

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Security and Privacy: Blockchains and distributed ledgers

Abstract

As quantum computers require highly specialized and stable environments to operate, expanding their capabilities within a single system presents significant technical challenges. By interconnecting multiple quantum processors, distributed quantum computing can facilitate the execution of more complex and larger-scale quantum algorithms. End-to-end heuristics for the distribution of quantum circuits have been developed so far. In this work, we derive an exact integer programming approach for the Distributed Quantum Circuit (DQC) problem, assuming fixed module allocations. Since every DQC algorithm necessarily yields a module allocation function, our formulation can be integrated with it as a post-processing step. This improves on the hypergraph partitioning formulation, which finds a module allocation function and an efficient distribution at once. We also show that a suboptimal heuristic to find good allocations can outperform previous methods. In particular, for quantum Fourier transform circuits, we conjecture from experiments that the optimal module allocation is the trivial one found by this method.

Problem

• Multi-Module Quantum Computer
• Quantum Circuit Optimization
• Task Allocation

Innovation

• Quantum Computing
• Mathematical Programming
• Task Allocation