简介:为非强迫的优化在方法上介绍研究。学习的假设;主要结果;在简化Armijo类型下面的方法的集中性质衬里搜索。
简介:Theconjugategradientmethodforunconstrainedoptimizationproblemsvarieswithascalar.Inthisnote,ageneralconditionconcerningthescalarisgiven,whichensurestheglobalconvergenceofthemethodinthecaseofstrongWolfelinesearches.ItisalsodiscussedhowtousetheresulttoobtaintheconvergenceofthefamousFletcher-Reeves,andPolak-Ribiere-Polyakconjugategradientmethods.Thattheconditioncannotberelaxedinsomesenseismentioned.
简介:在这份报纸,我们在存在上解决一个问题结合线性多步方法(LMSM)的symplecticity,否定结果被获得。[从作者抽象]
简介:Arevisedconjugategradientprojectionmethodfornonlinearinequalityconstrainedoptimizationproblemsisproposedinthepaper,sincethesearchdirectionisthecombinationoftheconjugateprojectiongradientandthequasi-Newtondirection.Ithastwomerits.Theoneisthattheamountofcomputationislowerbecausethegradientmatrixonlyneedstobecomputedonetimeateachiteration.Theotheristhatthealgorithmisofglobalconvergenceandlocallysuperlinearconvergencewithoutstrictcomplementaryconditionundersomemildassumptions.Inadditionthesearchdirectionisexplicit.
简介: