The introduction of five benchmark global
optimization test functions by De Jong has remained prominent in Mathematics
and Computer Science for over three decades now. This paper examines the effect
of the search population and the number of iterations of the Cuckoo Search and
the African Buffalo Optimization in providing solutions to one of Dejong
function, the Rosenbrock function, sometimes called Dejong2 function which is a
unimodal non-separable function. The Rosenbrock function because of its deceptive
flat landscape has proven to be a good test case for optimization algorithms
since the flat surface provides very misleading information to search agents.
After a number of experimental investigations using different iteration numbers
and population, this study concludes that the CS provides better solutions but
at a cost of more computer resources than the ABO. As a result, this study in
harmony with the No Free Lunch Theorem concludes that if speed is the main
consideration, the ABO is a better algorithm in solving the Rosenbrock (or a
similar function), otherwise, the CS is a better choice.
Primary Language | English |
---|---|
Journal Section | Articles |
Authors | |
Publication Date | August 19, 2018 |
Published in Issue | Year 2018Issue: 2 |