因子设计的重量计数器及其两种字长型的关系（A weight counter designed by a factor and its relation to two word lengths）.pdfVIP

因子设计的重量计数器及其两种字长型的关系 中文摘要 中文摘要 在实验设计中怎样评价设计的好坏是一个热点问题, 对于部分因子设计, 现有的文 献已经提出了多种准则, 其中最小低阶混杂准则（Minimum Aberration 简称MA）被广 泛地应用于具有定性因子正规部分因析设计中. 然而, 在具有定量因子的设计中, 仅仅 采用最小低阶混杂准则并不能刻画设计的统计特性. 比如, 因子的水平置换不改变设 计的广义字长型, 但却影响设计的统计推断能力. Cheng and Ye (2004) 在多项式模型 下定义了β-字长型来进一步比较含有定量因子的不同设计, 他们认为具有较小的β-字 长型的设计是比较好的. 在本文中我们通过定义一个设计D 的重量计数器E (D), 建 立了E (D) 与广义字长型的解析式, 以及E (D) 与β-字长型的解析式, 给出了计算β- 字长型的另一种算法. 进而推导出广义字长型与平均β- 字长型的关系表达式, 说明具 有较小的广义字长型的设计, 同时也会有比较小的平均β-字长型. 关键词: β-字长型; 因子设计; 广义字长型; 重量计数器 作 者: 王晓丽 指导老师: 唐 煜 I 英文摘要 因子设计的重量计数器及其两种字长型的关系 The evaluation of a design is a hot issue in the field of factorial designs. For frac- tional factorial designs, many kinds of criteria were proposed in some existing papers. Especially, the minimum aberration criterion is widely used in the selection of regular fractional factorial designs with qualitative factors. For designs with quantitative fac- tors, however, minimum aberration criterion can not describe statistical properties. For instance, although level permutation of factors could not alter the generalized word- length pattern, it could affect the ability of statistical inference of the design. Cheng and Ye (2004) defined the β-word-length pattern based on a polynomial model in or- der to compare the designs with quantitative factors. They argued that a design with less β-word-length pattern is recommended. In this thesis, we will define E (D) as the weight enumeration of the design D and give another method for computing β-word- length pattern. Then we can establish the relationship between