Emergence of spatial correlation of local processing and memory elements based on phase-change materials and its application to simulator for large-scale problems

2018 Fiscal Year Final Research Report

• Project/Area Number: 16H03889
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Single-year Grants
• Section: 一般
• Research Field: Optical engineering, Photon science
• Research Institution: Keio University
• Principal Investigator: Saiki Toshiharu 慶應義塾大学, 理工学部(矢上), 教授 (70261196)
• Project Period (FY): 2016-04-01 – 2019-03-31
• Keywords: 自然知能 / 組み合わせ最適化 / スピングラス / 相変化材料 / コロイド結晶

Outline of Final Research Achievements

We demonstrated two approaches to solve spin glass problem by implementing it into physical systems. The first approach is to replace the spin glass to a coupled oscillator system and to calculate its lowest eigenmode. We evaluated the algorithm by comparing with the conventional simulated annealing algorithm and found that the calculation time can be reduced by one order. We proposed to implement the algorithm into a coupled plasmon particles, where the strength of inter-particle interaction can be autonomously modified through dielectric screening using a phase-change material to come to the optimized solution efficiently. The second approach is to utilize a buckled phase of two-dimensional colloidal crystal, which is equivalent to the frustrated triangular spin lattice. We experimentally demonstrated frustration behavior by confining 1-um polystyrene beads in a two-dimensional slit and visualized the annealing process to reach the lowest frustrated solution.

Free Research Field

ナノ光学　光物性

Academic Significance and Societal Importance of the Research Achievements

相変化材料による演算・記憶機能が大域的に相関を形成することを実証し、その機構の多様性、多自由度性と複雑性に基づく入出力相関の中にコンピューティング機能、さらには広義の知性を見出す契機となった。これらの研究を通して非ノイマン型コンピューティングの新しい着想と設計論が生まれると期待され、工学的意義は大きい。
既存のアルゴリズムの実装によるコンピューティング機能の実証を通して、脳機能や複雑な自然・社会現象の理解を目的とした構成論的アプローチ、すなわち複雑な相互作用、相関形成機能を内在したシミュレータを実際に動かして現象の理解に役立てるへの応用という新しい視座を得ることができた。