講演情報
[9a-F212-5]Unconstrained Optimization of Experimental Parameters Using All-Valid-State HOBO for Fabrication of Atomic Junctions
〇Juncheng Wang1, Takumi Kanezashi1, Koki Awaya1, Haruya Nagata1, Qingxuan Lin1, Jun-ichi Shirakashi1, Tetsuo Shibuya2, Hiroshi Imai2 (1.Tokyo Univ. Agr. & Tech., 2.Tokyo Univ.)
キーワード：
Gate-Based Quantum Computer、Quantum-Classical Hybrid Algorithm、Variational Quantum Eigensolver
Integrating quantum computing with practical problems is an important goal. In our previous studies, we focused on the traveling salesman problem (TSP) and proposed the all-valid-state higher-order binary optimization (AVS-HOBO) model to repurpose invalid states generated by compact binary encoding. Here, we apply AVS-HOBO to feedback-controlled electromigration (FCE), which enables single-atom-level migration control. Since precise FCE requires selecting many experimental parameters during the experiment, the feedback voltage amount VFB (%) schedule is formulated as a combinatorial optimization problem. In AVS-HOBO encoding, invalid qubit states are cyclically mapped to valid VFB values, transforming the problem into an unconstrained optimization process. Using the variational quantum eigensolver (VQE), AVS-HOBO reduced the average residual energy from 2.07 × 10³ with QUBO encoding (90 qubits) to 9.68 × 10² with AVS-HOBO encoding (40 qubits).