包含纯净两因子交互作用的分辨率为Ⅳ的2^n-（n-k）设计的某些结果.pdfVIP

维普资讯 第 41卷 第 2期 南 开 大 学 学 报 (自然科学版) Vo1．41 N9-2 2008年 4月 ActaScientiarum Naturalium UniversitatisNankaiensis Apr．2008 ArticleID：0465—7942(2008)02—0080—05 SomeResultsonResolution Ⅳ rt--(以一五) DesignsContainingClearTwo——factorInteractions ZhaoShengli ．ZhangRunchu (1．SchoolofMathematicalSciences，NankaiUniversity，Tianjin300071，China； 2．SchoolofMathematicalSciences，QufuNormalUniversity，Qufu273165，China) Abstract：Cleareffectscriterionisoneofthemostoftenusedcriteriaforselectinggooddesigns．Thispaper isdevotedtobuildanupperboundonthemaximum numberofcleartwo—factorinteractionsin2一 “一’designs withresolution Ⅳ．Comparisonsshow thattheupperboundbehaveswellfork一 5，6，7，8，9． Keywords：fractionalfactorialdesign；clear；resolution CLC number：O212．6 Documentcode：A 0 Introduction By a2一“一 design，wemean atwo—levelfractionaldesignwith ，zfactorsand 2 runs，which isa 2一“一 fractionofafull2 designandiscompletelydecidedby，z——五independentdefiningwords．How to chooseagooddesign’iswhattheexperimentersconcern．Differentcriteriaarestudiedbymanystatisti— clan．M inimum aberrationandcleareffectscriteriaarethemostoftenusedtwocriteriaforselectinggood designs[’．W hentheexperimenter’sknowledgeisdiffuse。areasonableassumptionheorshecanmake isthehierarchicalassumption．Undersuchcircumstances，minimum aberrationisthemostoften used criterionforselectinggooddesigns． Differentsituationscallfordifferentdesigns．Asusual，weassumethatinteractionsinvolving three ormorefactorsarenegligible．A designofresolution V orhigherpermitstheestimationofallthemain effectsandtwo—factorinteractions(2fi’s)．Inwhatfollows，w