简介:本文提出了求解非线性方程组的一种非精确Broyden方法.该方法是文献[8]中精确Broyden方法的推广.在适当的条件下,我们证明了非精确Broyden方法具有全局收敛性和超线性收敛性.数值实验表明,该方法效果较好.
简介:AinteriorpointscalingprojectedreducedHessianmethodwithcombinationofnonmonotonicbacktrackingtechniqueandtrustregionstrategyfornonlinearequalityconstrainedoptimizationwithnonegativeconstraintonvariablesisproposed.Inordertodealwithlargeproblems,apairoftrustregionsubproblemsinhorizontalandverticalsubspacesisusedtoreplacethegeneralfulltrustregionsubproblem.Thehorizontaltrustregionsubprobleminthealgorithmisonlyageneraltrustregionsubproblemwhiletheverticaltrustregionsubproblemisdefinedbyaparametersizeoftheverticaldirectionsubjectonlytoanellipsoidalconstraint.Bothtrustregionstrategyandlinesearchtechniqueateachiterationswitchtoobtainingabacktrackingstepgeneratedbythetwotrustregionsubproblems.Byadoptingthel1penaltyfunctionasthemeritfunction,theglobalconvergenceandfastlocalconvergencerateoftheproposedalgorithmareestablishedundersomereasonableconditions.AnonmonotoniccriterionandthesecondordercorrectionstepareusedtoovercomeMaratoseffectandspeeduptheconvergenceprogressinsomeill-conditionedcases.
简介:Thispaperdiscussestheintervalestimationsmethodfortheparametersandotherreliabilitycharactersofathree-poxameterWeibulldistribution.Accordingtothefiducialdistrlbutiontheoryoftheparameter,theauthorpresentstheconfidenceintervalsoftheporameters,thereliabilityandthereliablelife.Anexamplemadsimulationresultsaregiven.Itisshownthatthemethodpresentedinthispaperispracticableandworthnoticing.
简介:本文致力于研究非线性中立型延迟积分微分方程隐式Euler方法的收缩性。本文中的Lipschitz数是关于变量t的函数,而不是常数,最终能得到其数值解的结果是收缩的。