A framework for testing scientific software : a case study of testing Amsterdam discrete dipole approximation software / by Hongbing Shao.

Author/creator Shao, Hongbing author.
Other author Ding, Junhua, degree supervisor.
Other author East Carolina University. Department of Computer Science.
Format Theses and dissertations
Publication[Greenville, N.C.] : [East Carolina University], 2016.
Description108 pages : illustrations (some color)
Supplemental ContentAccess via ScholarShip
Subjects

Summary Software testing with scientific software systems often suffers from test oracle problem, i.e., lack of test oracles. Amsterdam discrete dipole approximation code (ADDA) is a scientific software system that can be used to simulate light scattering of scatterers of various types. Testing of ADDA suffers from "test oracle problem". In this thesis work, I established a testing framework to test scientific software systems and evaluated this framework using ADDA as a case study. To test ADDA, I first used CMMIE code as the pseudo oracle to test ADDA in simulating light scattering of a homogeneous sphere scatterer. Comparable results were obtained between ADDA and CMMIE code. This validated ADDA for use with homogeneous sphere scatterers. Then I used experimental result obtained for light scattering of a homogeneous sphere to validate use of ADDA with sphere scatterers. ADDA produced light scattering simulation comparable to the experimentally measured result. This further validated the use of ADDA for simulating light scattering of sphere scatterers. Then I used metamorphic testing to generate test cases covering scatterers of various geometries, orientations, homogeneity or non-homogeneity. ADDA was tested under each of these test cases and all tests passed. The use of statistical analysis together with metamorphic testing is discussed as a future direction. In short, using ADDA as a case study, I established a testing framework, including use of pseudo oracles, experimental results and the metamorphic testing techniques to test scientific software systems that suffer from test oracle problems. Each of these techniques is necessary and contributes to the testing of the software under test.
General notePresented to the faculty of the Department of Computer Science.
General noteAdvisor: Junhua Ding.
General noteTitle from PDF t.p. (viewed January 30, 2017).
Dissertation noteM.S. East Carolina University 2016.
Bibliography noteIncludes bibliographical references.
Technical detailsSystem requirements: Adobe Reader.
Technical detailsMode of access: World Wide Web.

Availability

Library Location Call Number Status Item Actions
Electronic Resources Access Content Online ✔ Available