Disentanglement Provides a Unified Estimation for Quantum Entropies and Distance Measures
Myeongjin Shin, Seungwoo Lee, Junseo Lee, and 4 more authors
arXiv preprint 2401.07716, 2024
The estimation of fundamental properties in quantum information theory, including von Neumann entropy, Rényi entropy, Tsallis entropy, quantum relative entropy, trace distance, and fidelity, has received significant attention. While various algorithms exist for individual property estimation, a unified approach is lacking. This paper proposes a unified methodology using Layerwise Quantum Convolutional Neural Networks (LQCNN). Recent studies exploring parameterized quantum circuits for property estimation face challenges such as barren plateaus and complexity issues in large qubit states. In contrast, our work overcomes these challenges, avoiding barren plateaus and providing a practical solution for large qubit states. Our first contribution offers a mathematical proof that the LQCNN structure preserves fundamental properties. Furthermore, our second contribution analyzes the algorithm’s complexity, demonstrating its avoidance of barren plateaus through a structured local cost function.