City Research Online

Abstract

Group testing is a method for detecting the small number of important units from a large population in situations where groups of units may be tested together to observe whether any members of the group are important. If a group is found not to contain any important units then it is classified immediately, otherwise its members are investigated further. Group testing is also a special case of a search problem.

The fundamental ideas of search are discussed and formulated and group testing is expressed in terms of search notation. Links are made to other areas of search similar in nature to group testing.

A new binary group testing algorithm, multi-stage stepwise group testing is introduced and investigated. A study of this and other algorithms is carried out in terms of criteria other than the expected number of tests.

The case of additive group testing is considered and a new multi-stage algorithm is introduced. The expected number of tests of this algorithm is derived and its properties are compared to existing algorithms.

The problem of factor screening is discussed. It is argued that the only model assumption which has not been previously satisfactorily considered is that the error variance must be known. Two methods for dealing with this case are proposed, and their relative properties are investigated in a simulation study.

An application of group testing is presented using robust circuit design as an example, providing useful insight into the performance of group testing and providing a new area of application.

Publication Type:	Thesis (Doctoral)
Subjects:	Q Science > QA Mathematics T Technology > TA Engineering (General). Civil engineering (General)
Departments:	School of Science & Technology > Engineering School of Science & Technology > School of Science & Technology Doctoral Theses Doctoral Theses

Export

Downloads