【摘要】 <正> §1.引言 关于单因素优选法的术语见[1].所谓有(时间)延迟周期τ,是说当作第i次试验时只知道直到第i-τ-1次试验的结果.τ=1,2,3的情形分别在[2]和[3]中研究.[3]并对任意的τ,给出了n次(试验)策略可行区间长度L_n(τ)的上界A_n(τ).它在τ=2和3时可以达到.于是[3]猜测:对任何τ,L_n(τ)=A_n(τ).更多还原

【Abstract】 The minimax optimization of unimodal functions of one variable with time delay has practical value in the allocation of experiments that need a longer time interval to turn out their results. The optimal strategies suggested thus far are only for the periods not more than three. The upper bound A_n (τ) of the length of feasible interval was presented in [2] for any period of time delay. It is proved in this paper that, even when the period is four, the upper bound A_n (4) is irreachable. A pseudooptimal strategy approximate to the upper bound is suggestd, and its error is estimated By using the approach of this paper, it can be proved that A_n (5) is also irreachable.更多还原