2022 Fiscal Year Final Research Report
Development of interaction-based behavioral model using quantum computing technique
Project/Area Number |
20K14844
|
Research Category |
Grant-in-Aid for Early-Career Scientists
|
Allocation Type | Multi-year Fund |
Review Section |
Basic Section 22050:Civil engineering plan and transportation engineering-related
|
Research Institution | The University of Tokyo |
Principal Investigator |
Urata Junji 東京大学, 大学院工学系研究科(工学部), 講師 (70771286)
|
Project Period (FY) |
2020-04-01 – 2023-03-31
|
Keywords | 行動モデル / 相互作用 / 量子コンピューティング |
Outline of Final Research Achievements |
We developed a method for computing simulation runs under the interaction model as a n-body simultaneous choice problem. The problem was that the number of candidate alternatives explodes in combination when the simulation execution is considered as a simultaneous choice problem, but we transformed the equation to a QUBO-type equation that can be applied to a quantum computer and made it possible to solve the problem by quantum computation. Numerical calculations on a quantum computer can be performed in ms even when the number of decision makers is 100, and the cost of simulation calculation has been successfully reduced to the limit. We also compared the exact solution obtained with the exact solution obtained on a scale that allows full enumeration calculations, and confirmed that the proposed algorithm obtained an exact solution.
|
Free Research Field |
交通工学
|
Academic Significance and Societal Importance of the Research Achievements |
相互作用下の同時選択問題をごく短時間で解くアルゴリズムは、多くの予測シミュレーションに実装される可能性がある。これまで、同時選択問題を解けないため、エージェントの相互作用を逐次手番型で解くという工夫がほとんどのシミュレーションで実装されていたが、置き換えることが可能になる。逐次手番型であれば解くことはできても、エージェント数分の繰り返し計算が必要であり、計算コストの縮減には限界があり、エージェント群に働きかける政策最適化などの計算には限界があったが、本研究により解決可能となる。例えば、自動運転・手動運転の混在時の最適な自動運転車の制御などにも適用可能と考える。
|