Dr. Megan Olsen from Loyola University Maryland will present at the Seminar Friday April 13 at 1:15pm in Room 121 of Science Hall (SH). Dr. Olsen’s presentation title, abstract, and biographical sketch are below.
Title: Metamorphic Testing for Simulation Validation
Author: Dr. Megan Olsen
Location: Science Hall (SH http://www.uta.edu/maps/?building=SH) Room 121
Date: Friday, April 13
Time: 1:15pm – 2:15pm
Abstract: A simulation model must be validated, i.e. demonstrated to accurately represent the system it studies, for its results to be reliable. However, simulation validation is a challenging task. One of the primary sources of this challenge is the absence of an oracle to test its validity, as without an oracle one cannot determine if the result of a program is correct. This oracle problem has long been studied in software engineering. Metamorphic Testing has been found to be an effective technique to test software without an oracle by creating pairs of test cases that act as pseudo-oracles. We provide guidelines on applying a modified version of metamorphic testing to increase confidence in the validity of simulation models. In this technique, pseudo-oracles are developed based on metamorphic relations between parameters and behaviors within an executable simulation model. These relations represent an understood property of the system being studied, which can be used to check the validity of the simulation without knowing the correct answer. In this talk we will explain metamorphic testing and how it applies to simulation model validation, including a demonstration of its effectiveness via a case study on a gossip propagation model.
Biographical Sketch: Dr. Megan Olsen is an Associate Professor of Computer Science at Loyola University Maryland. She earned her M.S. (2009) and Ph.D. (2011) in computer science from University of Massachusetts Amherst, and her B.S. in computer science from Virginia Tech (2005). Dr. Olsen’s research currently focuses on improving simulation approaches and validation. Recent work includes quantifying the level of validation achieved on a simulation model, guidelines for using metamorphic testing for simulation validation, and utilizing reinforcement learning within agent-based predator-prey models. Recent work has been published in Summer Sim, Winter Sim, Spring Sim, and the International Conference on Computational Science.