Binary crow search algorithm for the uncapacitated facility location problem

Abstract

The crow search algorithm (CSA) is a recently proposed population-based optimization algorithm for continuous optimization. Since the original CSA searches for a feasible solution in a continuous search space, it cannot handle binary optimization problems directly. A few binary variants of CSA are presented in the literature. However, these variants search for a new solution in the continuous domain and need transfer functions to adapt the solution to the binary domain. This may cause poor exploration, making some regions in the search space impossible to discover. This paper proposes an effective binary CSA (BinCSA) using bitwise operations that directly searches for a feasible solution in the binary search space. For this purpose, the original update mechanism of the CSA is improved using exclusive-OR and AND logical operators in order to provide a good balance between exploration and exploitation in the binary search space. The effectiveness of the proposed BinCSA is evaluated on the uncapacitated facility location problem (UFLP), one of the most widely investigated pure binary optimization problems. The performance of BinCSA is examined using two different UFLP datasets, ORLIB and M*. The experimental results show that BinCSA obtained the optimal solution for 13 out of 15 instances of ORLIB and 12 out of 20 instances of M*. Moreover, BinCSA exhibits superior performance on ORLIB instances when compared to other methods and is very competitive on M* instances in terms of solution quality and robustness. The source code for BinCSA, as used for the UFLP, is available at https://github.com/3mrullah/BinCSA.