Performance Evaluation of Genetic Algorithm and Simulated Annealing in Solving Kirkman Schoolgirl Problem.

This paper provides performance evaluation of Genetic Algorithm and Simulated Annealing in view of their software complexity and Simulation runtime. Kirkman Schoolgirl is about arranging fifteen schoolgirls into five triplets in a week with a distinct constraint of no two schoolgirl must walk together in a week. The developed model was simulated using Matlab version R2015a. The performance evaluation of both Genetic algorithm and Simulated Annealing was carried out in terms of program size, program volume, program effort and the intelligent content of the program. The results obtained show that the runtime for GA and SA are 11.23sec and 6.20sec respectively. The program size for GA and SA are 2.01kb and 2.21kb, respectively. The lines of code for GA and SA are 324 and 404, respectively. The program volume for GA and SA are 1121.58 and 3127.92, respectively. The program effort for GA and SA are 135021.70 and 30633.26 respectively, while the intelligent content of the program for GA and SA are 72.461 and 41.06, respectively. Both Algorithms are good solvers, however it can be concluded that Genetic Algorithm outperformed simulated Annealing in most of the evaluated parameters. Keywords: Genetic Algorithm, Simulated Annealing, Kirkman Schoolgirl, software complexity and simulation runtime