Presentation Information
[9a-F212-6]Variational Quantum Eigensolver with Conditional Value at Risk for Solving Traveling Salesman Problem
〇Qingxuan Lin1, Juncheng Wang1, Takumi Kanezashi1, Haruya Nagata1, Koki Awaya1, Jun-ichi Shirakashi1, Tetsuo Shibuya2, Hiroshi Imai2 (1.Tokyo Univ. Agr. & Tech., 2.Univ. Tokyo)
Keywords:
Variational Quantum Eigensolver,Conditional Value at Risk,Traveling Salesman Problem
The Traveling Salesman Problem (TSP) is a representative NP-hard combinatorial optimization problem. In previous work, we investigate the application of the Variational Quantum Eigensolver (VQE) with a Higher-Order Binary Optimization (HOBO) formulation for solving TSP on NISQ-era quantum devices. Conventional VQE evaluates the objective function using the expectation value of all measurement outcomes, which can reduce optimization performance due to the influence of high-energy solutions.To address this issue, we introduce the Conditional Value at Risk (CVaR) objective. Instead of averaging over all sampled solutions, CVaR focuses on the lowest-energy fraction of measurement outcomes, thereby emphasizing high-quality candidate solutions during optimization. By reducing the impact of unfavorable samples, the proposed CVaR-VQE framework aims to improve the effectiveness of quantum variational optimization for TSP and other combinatorial optimization problems.