Shuttling Compiler for Trapped-Ion Quantum Computers Based on Large Language Models
Researchers have developed a novel compiler for trapped-ion quantum computers that leverages large language models (LLMs). This LLM-based compiler is designed to generate efficient schedules for qubit shuttling, a critical operation in quantum computing. By fine-tuning pre-trained LLMs on various shuttling architectures, the system demonstrates an ability to generalize to unseen layouts and outperforms existing compilers, reducing shuttling effort by up to 15% on benchmark circuits. AI