A Study of the Effects of the Population on a Genetic Algorithm
This book is the documentation of the experiment I conducted, attempting to determine whether or not the number in the initial population of a genetic algorithm drastically affected the productivity of such a genetic algorithm. More
This experiment was conducted to better understand how the number in the initial population of a genetic algorithm affects its effectiveness. The experiment was conducted in this procedure. First, open the genetic algorithm program with an adjustable population. Second, enter the number into the program that is your current independent variable. Third, press the "return" key to start the calculations. Fourth, wait until the program stops displaying new data. Fifth, record the last generation shown in the console. Sixth, Reiterate upon the steps above excluding step 1 to conduct all trials. The data is as follows in the format, number in the population : generations. 10 : 86, 94, 120, 91, 58, 66, 79, 51, 44, and 88; 15 : 32, 44, 60, 51, 55, 44, 45, 33, 44, & 33; 20 : 51, 53, 22, 66, 48, 31, 39, & 33; 25 : 17, 31, 25, 34, 22, 32, 43, 23, & 24; and 30 : 27, 19, 29, 35, 34, 26, 29, 25, 18, & 31. The data that resulted from following this procedure was definitive as to a positive correlation between the independent and dependent variables, but showed smaller changes every other change in the independent variable. This could be attributed towards the computer processing unit slowing down with the increased workload and not being able to process as efficiently. Building a new program to better access the multithreading capabilities of a modern central processing unit would decrease this slowdown as would running the program on a supercomputer or potentially a quantum computer.
Available ebook formats: