Using automated search to generate test data for matlab

The critical functionality of many software applications relies on code that performs mathematically complex computations. However, such code is often difficult to test owing to the compound datatypes used and complicated mathematical operations performed. This paper proposes the use of automated search as an efficient means of generating test data for this type of software. Taking Matlab as an example of widely-used mathematical software, a technical framework is described that extends previous work on search-based test data generation in order to handle matrix datatypes and associated relational operators. An empirical evaluation demonstrates the feasibility of this approach. Categories and Subject Descriptors D.2.4 [Software Engineering]: Software/Program Verification; I.2.2 [Artificial Intelligence]: Automatic Programming; G.4 [Mathematics of Computing]: Mathe

• Type: Conference
• Year: 2009
• Where: GECCO
• Authors: Sion Ll Rhys, Simon M. Poulding, John A. Clark
• Keywords: Artificial Intelligence, Compound Datatypes, GECCO 2009, Test Data, Widely-used Mathematical Software