GA-hardnessandinterdependencebetweengenesinthechromosomeareimportantquestionsinthestudyofgeneticalgorithms(GA).Traditionalmethods,whichareusedtomeasuretheinteractionbetweengenes,canonlyreflecttheextentofepistasisbetweenallgenesinthechromosome.Therefore,thedefinitionofthefitnesslandscapeofschemataisproposedinthispaper,andepistasismeasuresonthislandscapeofschemataareusedtoanalyzethedegreeofinterdependencebetweensomecertaingenelociinstudy.Someinformationbetweenthesesitescanbereflectedbysomecharactersofthefitnesslandscapeofschematawhicharecomposedofthesefixedsites.Thestrongertheinteractionbetweenthesesites,thelargerthevariationofthefitnessofschematawhosefixedsitescorrespondtothosesitesinstudy,andthemoreruggedthefitnesslandscapeoftheseschemata.Accordingtothedegreeofinteractionbetweenthesegivengeneloci,buildingblocksofGAcanbeanalyzedanddetermined,andfurthergeneticoperatorsandthestructureofGAcanbedesignedandadjustedtoimprovetheperformanceofGA.Atlast,alotofexperimentsincludingNK-modelsaredone,andresultsofempiricalanalysisshowthatthismethodiseffective.
