約束MINIMAX問(wèn)題的廣義單調(diào)線搜索SQP算法.pdf

1、廣西大學(xué)碩士學(xué)位論文約束MINIMAX問(wèn)題的廣義單調(diào)線搜索SQP算法姓名：張雪露申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：應(yīng)用數(shù)學(xué)指導(dǎo)教師：簡(jiǎn)金寶20060501GENERALIZEDMoNOTONELINESEARCHSQPALGoRITHMFoRCoNSTRAINEDMINIMAXPROBLEMSABSTRACTIti8wenknownthatthea190rith腳of卵quentialquadraticprogramming(SQP)areone

2、ofthe螂tcl鵝8icalalgorith瑚for801vjngnonlinearcon8traintedoptimizationproble腳，andtheygeneraIlyha代goodconvergenceproperti姻，刪mericalexperimentsal∞showthatSQPalgorith腳areverye骶ctiveHo啪ver，m蹦toftheSQPalgDrithmsarefacedwithanimp

3、ortantprobI咖，thatis，ateachite確越onthe艙algorithⅡ磚requirethatthequadraticprogramming(QP)subprobIe嘲ml埽tbe80lvableDI坶tothecDnstrajntsofQPrepre孵nta五礙t—ordetlinearizationoforiginalproblem’scon8traints，80Ⅱ時(shí)fe柏|bk8酏ofQPm硒rbeempty

4、Inordertoovercomethe8hortcoming，thereappear80menewtechⅡiq懈，州ch柚pivoting印eration，iⅡ‘roducingnewvariabl髑qbobtainsuperlirlearocnvergenceand刪dMaratos疵t，0ne8houldB01veoneornIore8ubprogrammin鏟togetasecondordercorrectjondirecti

5、on，thi8kadstol盯ge8caI船ofcomputa土io珊Inthi8paper，the∞nlinearmini瑚xprobleⅡ坶withgen盱缸c0璐traint8aredi8c118sedBymea璐0fpivotingoperationWec唧tructoneQPBubproblemtogetanimproVeddirectionateachiterationSimnarly岫ingthe咖tructionalpr

6、incipleofQP8ubprobkmweonlyneedomsyBtemsofLinearEq眥io娜(sLE)toobtainaoorrectiondirectionInconnectionwitha印ecialmeritfunctiontheg∞eralizedmonotoneU啪8earchi8llsedtoyieldthe8t印sizeSoanew訕0ritlllnfbrBolvingthedisclls∞dproblel瑚

7、培pr郫entedUndermiIdconditio璐，wecanobtaintheglobal姐d8uperlinear咖rgence鼬Ⅱlarbblywhikadding80nle∞mputatiotl8about印proximatelyactivec0璐traiⅡt8ub∞t，buttheco瑚traint8numberofsubprobk腳is刪chl嘲thanbe￡oreFin缸ly80me肌mericale)【perimen