A TWO-ARMED BANDIT PROBLEM WITH ONE ARM KNOWN UNDER SOME CONSTRAINTS

スポンサーリンク

概要

• 論文の詳細を見る

• A finite-time horizon two-armed bandit problem with one arm known, in which there are constraints on the number of times each arm may be pulled, is considered. The loss incurred at the stage when there are n stages remaining is multiplied by the factor β_n, and the objective is to minimize the total expected weighted loss over n stages. This problem is formulated by dynamic programming and the critical value function which specifies the optimal strategy is derived along with its monotonicity properties. The recursive equation is solved explicitly in the case of the exponential distribution. Tables of the critical values are obtained for special cases of {β_n}.

• 社団法人日本オペレーションズ・リサーチ学会の論文

著者

• Hamada Toshio

  Himeji College

• Hamada Toshio

  Himeji College Of Hyogo

• Ross Sheldon

  University of California

• Ross Sheldon

  University Of California Berkeley

関連論文

• A TWO-ARMED BANDIT PROBLEM WITH ONE ARM KNOWN UNDER SOME CONSTRAINTS
• Analyzing Methods for Production Logistics by Orthonormal System : 3rd Report, The Case of Considering Processing of Parts and Assembling
• A TWO-ARMED BANDIT PROBLEM WITH ONE ARM KNOWN UNDER SOME CONSTRAINTS

スポンサーリンク