Abstract: Sine Cosine Algorithm (SCA) is one of the highly-referred optimization algorithms in the literature. The present study contributes by discovering three shortcomings of SCA and making a few recommendations. We show that the mathematical model of SCA does not work as explained in the original paper and its performance can be improved by modifying the position-updating equation of SCA as stated in the original paper. Moreover, we empirically and statistically show that sine and cosine functions, which make this algorithm different from the others, can be replaced with a simple random variable having a value in the range [−1, 1] without degrading the overall performance of the algorithm. Furthermore, we demonstrate that the behavior of SCA is biased for the functions having the global optimum at the origin. Finally, on the basis of the analysis of SCA, we make two recommendations for the metaheuristics designers regarding the selection of the benchmarks and mapping of the inspiration when designing a new algorithm.
- Mathematical model of SCA does not work as explained in the original paper.
- Equivalent performance can be obtained by replacing sine cosine functions with aa random variable.
- The behavior of SCA is biased for the problems having global optimum at the origin.