Noise-Directed Warm-Starting Boosts 100-Qubit Quantum Optimization

Researchers at USRA, The Hartree Centre, and IBM Quantum have demonstrated a high-performance quantum optimization approach leveraging 100-qubit Ising Hamiltonians, achieving results comparable to the highest quality at that scale with similar algorithms. The team reports implementing Noise-Directed Adaptive Warm-Starting (ND-AWS), which exploits naturally occurring amplitude-damping-like noise components through bitflip gauge transformations, rather than attempting to correct for them. “The simplicity of the framework opens the door for future enhancements such as adaptive bias schedules, and integration with classical solvers,” the researchers write. ND-AWS improves performance over existing iterative Warm-Starting methods without requiring any additional circuit complexity, suggesting a path toward more readily implementable advances in quantum optimization. Results showed approximation ratios ranging from 0.974 to 1.0 for Erdős-Rényi graphs with 10% edge probability, to 0.989 to 1.0 for regular graphs. Demonstrations of quantum optimization are now tested at…