ISING MODEL GENERATION SYSTEM FOR CONSTRAINT, COMBINATION OPTIMIZATION CALCULATION SYSTEM AND ISING MODEL CALCULATION METHOD FOR CONSTRAINT

• Main Authors: WATANABE MASAHIRO, OKUYAMA TAKUYA, KATO DAICHI, KURODA EISUKE, KIRIHARA KENTA
• Format: Patent
• Language: English; Japanese
• Published: 27.12.2021
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

To preserve constraint satisfaction when a binary expression higher than a third order is converted into an Ising model, in order to solve a combination optimization problem.SOLUTION: A constraint satisfaction identification number list is generated by assigning identification numbers to all states of binary data to serve as constraint. A pseudo Ising model for identification is structured by giving auxiliary variables to identification numbers of combinations satisfying the constraint. A pseudo Ising model for mapping, in which a value by a combination of determination variables matches the identification number, is structured. A determination variable selection pseudo Ising model is structured by squaring a difference between the pseudo Ising model for identification and the pseudo Ising model for mapping. An auxiliary variable selection pseudo Ising model for selecting only one of the auxiliary variables given to the identification numbers is structured. A pseudo Ising model for constraint is structured by adding a penalty factor to each of the determination variable selection pseudo Ising model and the auxiliary selection Ising model and adding them together.SELECTED DRAWING: Figure 1 【課題】組合せ最適化問題を解くために、三次以上の二値式をイジングモデルに変換する際に制約充足性を保存する。【解決手段】制約となる二値式データのすべての状態に識別番号を割り当てて制約充足識別番号リストを生成し、制約を満たす組合せの識別番号にそれぞれ補助変数を与えることで識別用疑似イジングモデルを構築し、決定変数の組合せによる値が識別番号と一致するマッピング用疑似イジングモデルを構築し、識別用疑似イジングモデルとマッピング用疑似イジングモデルとの差分を二乗することで決定変数選択疑似イジングモデルを構築し、識別番号に与えられた補助変数を一つのみ選択する補助変数選択疑似イジングモデルを構築し、決定変数選択疑似イジングモデルと補助選択イジングモデルにそれぞれペナルティ係数を与えて足し合わせることで制約用疑似イジングモデルを構築する。【選択図】図１