Analyzing of Usage Effect of the Distribution Functions for SMDO Algorithm via Benchmark Function with Matlab Toolbox

Özet

This paper presents solution comparisons of benchmark functions by using stochastic multi-parameters divergence (SMDO) method with using different distribution functions. Using benchmark functions is an important method in measuring the effectiveness of algorithms. Because benchmark functions are used by all algorithm producers while trying their algorithms and this provides a good tool for the others to compare their algorithms with similar procedures. Benchmark functions is used in this paper for the main purpose of analyzing randomization process. It is known that distribution functions take place a vital role in getting random numbers. These random numbers are used in stochastic methods through specifying step size. It is believed that a suitable random number generation can support the search processes of algorithms. In this study the effects of distribution functions on benchmark functions are analyzed. For this purpose, a program is developed with MATLAB. The comparisons via the help of this program is shown in tabular form. The results are analyzed from the viewpoint of whether developing the randomization process makes contribution to problem solving power of algorithms. In this study SMDO algorithm is analyzed with different distribution functions by using different benchmark functions. In addition, in the study, a useful friend-friendly Matlab toolbox is proposed in which SMDO algorithm can be tested over different benchmark functions according to different distribution functions. (https://www.mathworks.com/matlabcentral/fileexchange/75044-smdo-with-distribution-function-for-benchmarking)

Abstract

This paper presents solution comparisons of benchmark functions by using stochastic multi-parametersdivergence (SMDO) method with different distribution functions. Using benchmark functions is animportant method in measuring the effectiveness of algorithms. Because benchmark functions are used byall algorithm producers while trying their algorithms and this provides a good tool for the others to comparetheir algorithms with similar procedures. Benchmark functions are used in this paper for the main purposeof analyzing randomization process. It is known that distribution functions take place a vital role in gettingrandom numbers. These random numbers are used in stochastic methods through specifying step size. Itis believed that a suitable random number acquisition process can support the search processes ofalgorithms. In this study the effects of distribution functions on benchmark functions are analyzed. Forthis purpose, a program is developed with MATLAB. The comparisons via the help of this program isshown in tabular form. The results are analyzed from the viewpoint of whether developing therandomization process makes contribution to problem solving power of algorithms. In this study SMDOalgorithm is analyzed with different distribution functions by using different benchmark functions. Inaddition, in the study, a useful friend-friendly Matlab toolbox is proposed in which SMDO algorithm canbe tested over different benchmark functions according to different distribution functions.(https://www.mathworks.com/matlabcentral/fileexchange/75044-smdo-with-distribution-function-forbenchmarking)