analiza aiqbench
This paper applies multi-agent LLMs to quantum compilation, the step that maps programs onto specific hardware, where it reports runtime and fidelity gains over the PowerMove baseline. The main claim to check is the benchmark comparison on "realistic workloads," which depends on hardware models and workload choices the abstract doesn't specify. The code is public, so readers can verify the results and see whether the LLM-discovered shuttling strategies carry over to other architectures.
Qubrio: High-Performance Quantum Compilation via Multi-Agent LLM Collaboration
arXiv:2611.02269v1 Announce Type: cross Abstract: Large Language Models (LLMs) offer an adaptable alternative to traditional quantum compilation heuristics, which bottleneck progress by demanding costly manual redesigns whenever hardware evolves. However, naive single-pass LLM approaches fail under the unique challenges introduced by quantum compilation. We introduce Qubrio, a practical agentic framework that significantly reduces heuristic redesigns through a three-level decomposition. By (1) partitioning programs into sequential operation stages, (2) deploying specialized agents for placement, routing and optimization, and (3) using disaggregated feedback for precise error correction, our system reliably navigates complex physical constraints. Beyond serving as a direct compiler, our LLM framework also uncovers novel strategies that facilitate convoy-style shuttling, substantially enhancing hardware concurrency without sacrificing fidelity. We further integrate these strategies into the previous state-of-the-art (SOTA) compiler, recovering a significant fraction of the LLM's performance advantages. On realistic workloads, our compiler achieves significant improvements in hardware runtime and program fidelity over the SOTA baseline PowerMove, seamlessly adapting to new hardware capabilities. Qubrio features an open user interface for practical use, and its code is publicly available at https://github.com/crosaa/Qubrio.
Authors:Jixuan Ruan, Zhuo Cui, Zhengding Hu, Zhongkai Yu, Xiang Fang, Yue Guan, Jason Ludmir, Phillip E. Weinberg, Xiuzhe Luo, Shengtao Wang, Sonia Lopez Alarcon, Hezi Zhang, Yufei Ding
Abstract:Large Language Models (LLMs) offer an adaptable alternative to traditional quantum compilation heuristics, which bottleneck progress by demanding costly manual redesigns whenever hardware evolves. However, naive single-pass LLM approaches fail under the unique challenges introduced by quantum compilation. We introduce Qubrio, a practical agentic framework that significantly reduces heuristic redesigns through a three-level decomposition. By (1) partitioning programs into sequential operation stages, (2) deploying specialized agents for placement, routing and optimization, and (3) using disaggregated feedback for precise error correction, our system reliably navigates complex physical constraints. Beyond serving as a direct compiler, our LLM framework also uncovers novel strategies that facilitate convoy-style shuttling, substantially enhancing hardware concurrency without sacrificing fidelity. We further integrate these strategies into the previous state-of-the-art (SOTA) compiler, recovering a significant fraction of the LLM's performance advantages. On realistic workloads, our compiler achieves significant improvements in hardware runtime and program fidelity over the SOTA baseline PowerMove, seamlessly adapting to new hardware capabilities. Qubrio features an open user interface for practical use, and its code is publicly available at this https URL.
Quantum Physics (quant-ph); Artificial Intelligence (cs.AI)
arXiv:2611.02269 [quant-ph]
(or arXiv:2611.02269v1 [quant-ph] for this version)
https://doi.org/10.48550/arXiv.2611.02269 arXiv-issued DOI via DataCite (pending registration)
Text integral sub licența deschisă a publicației · 224 cuvinte