Magic-Informed Quantum Architecture Search
Researchers propose a magic-informed quantum architecture search (QAS) technique that controls nonstabilizerness (magic) in quantum circuit design. Inspired by AlphaGo, the method uses Monte Carlo Tree Search with a Graph Neural Network (GNN) to estimate magic, steering searches toward high- or low-magic regimes. Benchmarked on ground-state energy and quantum state approximation problems, the technique effectively influences magic across the search tree.
Key facts
- Nonstabilizerness (magic) is a fundamental resource for quantum advantage.
- The technique enables control over magic in circuit design.
- Inspired by the AlphaGo approach.
- Uses Monte Carlo Tree Search with a Graph Neural Network (GNN).
- GNN estimates magic of candidate circuits.
- Steers search toward high- or low-magic regimes.
- Benchmarked on ground-state energy and quantum state approximation problems.
- Results show effective influence on magic across the search tree.
Entities
—