Probing entanglement and parameter sensitivity in QAOA via Quantum Fisher Information
Peer-Reviewed Publication
Updates every hour. Last Updated: 12-May-2026 00:15 ET (12-May-2026 04:15 GMT/UTC)
This article investigates Quantum Fisher Information (QFI) as a diagnostic tool for analyzing parameter sensitivity and entanglement in the Quantum Approximate Optimization Algorithm (QAOA).
Key Findings
Problem Analysis: The study examines Max-Cut problems on cyclic and complete graphs, plus random Ising model instances, comparing RX-only and hybrid RX-RY mixers up to depth p=9 .
QFI Insights: Complete-graph Max-Cut instances generate substantially larger QFI eigenvalues than cyclic ones, exceeding shot-noise scaling (4N) while remaining below the Heisenberg limit (4N2).
Entanglement Effects: The first entangling stage produces the dominant QFI increase, while additional stages yield diminishing returns. Entanglement primarily amplifies cross-parameter correlations rather than individual parameter sensitivity.
Practical Application: The authors propose QFI-Informed Mutation (QIm), a heuristic that adapts mutation probabilities using diagonal QFI entries. QIm outperforms uniform and random-restart baselines, especially for deeper circuits.
The work positions QFI as both a structural probe and practical optimization resource for NISQ-era quantum algorithms.