AI and the World's Hard Problems · Quantum physics & quantum computing · arXiv · 2026-10-19
Leveraging verifiable structure in quantum circuits to find more efficient compilations with agentic AI
The paper proposes using AI agents to compile quantum circuits more efficiently when bottleneck subroutines have classically verifiable structure, which allows approximate synthesis to be certified by fuzz testing. It discusses three applications: computing 1/√x for Coulomb operator simulation in fusion reactor modeling, decoding Reed-Solomon codes in decoded quantum interferometry, and quantum lookup tables with Givens rotations in chemistry simulation. Reference implementations are presented as the basis for a planned Kaggle competition on AI-driven compilation optimization.