Reproduce “Policy Gradient based Quantum Approximate Optimization Algorithm” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jiahao Yao, Marin Bukov, Lin Lin
A team at UC Berkeley and LBNL introduced PG-QAOA, a reinforcement learning framework that robustly optimizes Quantum Approximate Optimization Algorithm parameters for noisy intermediate-scale quantum devices. This derivative-free method effectively navigates complex control landscapes while being resilient to Hamiltonian, measurement, and classical noise.
Key insights
- Optimizing a probability distribution over QAOA parameters naturally provides robustness against diverse noise types and enables navigation of complex control landscapes with multiple optimal solutions.
- Policy-gradient-based reinforcement learning is an effective and compatible approach for quantum control on NISQ devices, as it operates without explicit derivative information or models of unknown noise sources.
- The width of the learned parameter distribution can serve as a qualitative indicator for the magnitude of unknown noise, offering a form of inherent error monitoring.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.