Exécution de notre outil de synthèse sur un nouvel article
Vérifiez l’article ci-dessous, puis lancez sa synthèse.
C-3PQ: A Closeness Centrality-based Circuit Partitioner for Quantum Simulations
Doru Thom Popovici (Lawrence Berkeley National Laboratory), Harlin Lee (University of North Carolina at Chapel Hill), Naoki Yoshioka (RIKEN), Mauro Del Ben (Lawrence Berkeley National Laboratory), Nobuyasu Ito (RIKEN), Katherine Klymko (Lawrence Berkeley National Laboratory), Daan Camps (Lawrence Berkeley National Laboratory), Anastasiia Butko (Lawrence Berkeley National Laboratory)
Résumé : Simulating quantum circuits (QC) on high-performance computing (HPC) systems is essential for validating quantum algorithms and probing the potential of large-scale quantum computation before the advent of scalable quantum hardware. However, state vector simulations are resource-intensive, often requiring large clusters with thousands of compute nodes and large amounts of memory. To address this, we introduce C-3PQ, an end-to-end framework that minimizes inter-node data movement through an efficient circuit partitioning scheme alongside a flexible code generator to offer a portable simulation solution. By formulating the distribution of the state vector and circuit over multiple nodes as a graph problem, we utilize closeness centrality to assess gate importance and design a fast, scalable partitioning method. C-3PQ compiles the resulting partitions into highly optimized kernels targeting a wide range of hardware platforms from Intel, AMD, NVIDIA and ARM. Moreover, compared to a state-of-the-art implementation specific for NVIDIA platforms, C-3PQ achieves a speedup of up to \(40\%\) at a fraction of the cost when partitioning the state vector.