变分不等式问题的次梯度外梯度投影算法研究-运筹学与控制论专业论文 word格式.docxVIP

性结果，只证明了算法的迭代序列{xk}是有界的.在本章中，我们给出了该算法的全局收敛性证明，弥补了该算法的缺憾.关键词变分不等式问题:次梯度外梯度投影算法;Lipschitz连续;类Armijo步长OnSubgradient Extragradient Projection AlgorithmsforSolvingVariational Inequality ProblemsABSTRACTVariational inequality problem(VIP) isaveryimportant research field intheOperational Research andhas awide range of applications in signalprocessing，image reconstruction，system identification，automatic control and other science且elds. Ithasbeen concerned widelybymanydomesticandforeignexpertsandscholarssincebeingproposed.Some 仕uitfulresultshavebeenachieved in itstheoriesandalgorithms.Among others，theprojection-tye algorithms areimpor-tant andrepresentative in these algorithms. Itiswell known，if theprojectioniseasilycalculated，projection-typealgorithmsarenotonlysimplein form，butalsopracticalandeffective，smallspaceoccupied，suitablefordealingwithlargescaleproblems.However，insomecasesitisimpossibleorneedstoomuchworktoexactlycomputetheorthogonalprojection，whichwillcertainlyaffecttheco且ver-gencespeed. Toovercome this shortcoming of the projection algorithm，Censorpresented thesubgradient extragradient projection algorithm. Tosomeexte旧，hehasreduced thecomputationalcomplexity ofthealgorithm. Onthisbasis，wegiveafurtherstudyforthesubgradientextragradient projectionalgorithmin thisarticle. Thefullthesisisdividedintothreechapters.The且rst chapter isa丑introduction.We describe the application back-groundsandtheresearchsituationsofthevariationalinequalityproblems(VIP)andthemainresultsobtainedinthisarticle.Inthe second chapter，we proposetwo Armijo-like stepsize subgradient ex-tragradientprojectionalgorithmsandtheirinexactformsforsolvingvariationalinequalityproblems.Inordertoovercomethedifficultyonthecomputationoftheorthogonalprojection，Censorproposesaclassofsubgradientextragradientprojectionalgorithms，whichreplacethe closedconvexsetwithaspecialstructurehalf-space. Intheconverge丑ceanalysis，heneedstoassumethatthemapFisLipschitzcontinuous.Toremovethisstrongcondition，weimprovethesubgrad