简介：LetXt(x)bethesolutionofstochasticdierentialequationswithsmoothandboundedderivativescoeffcients.LetXnt(x)betheEulerdiscretizationschemeofSDEswithstep2-n.Inthisnote,weprovethatforanyR>0andγ∈(0,1/2),supt∈[0,1],|x|≤R|Xnt(x,ω)-Xt(x,ω)|≤ξR,γ(ω)2-nγ,n≥1,q.e.,whereξR,γ(ω)isquasi-everywherefinite.

简介：InthispapersufficientconditionsformeanconvergenceandrateofconvergenceofHermite-FejertypeinterpolationintheLpnormonanarbitrarysystemofnodesarepresented.

简介：ＣＯＭＰＬＥＴＥＣＯＮＶＥＲＧＥＮＣＥＯＦＥＲＲＯＲＶＡＲＩＡＮＣＥＥＳＩＴＩＭＡＴＥＳＵＮＤＥＲ￠－ＭＩＸＩＮＧＥＲＲＯＲＩＮＬＩＮＥＡＲＭＯＤＥＬＳＸｕｅＬｉｕｇｅｎ（薛留根）（ＸｕｃｈａｎｇＴｅａｃｈｅｒｓＣｏｌｌｅｇｅ，Ｈｅｎａｎ４６１０００...

简介：Afewweakandstrongconvergencetheoremsofthemodifiedthree-stepiterativesequencewitherrorsandthemodifiedIshikawaiterativesequencewitherrorsforasymptoticallynon-expansivemappingsinanynon-emptyclosedconvexsubsetsofuniformlyconvexBanachspacesareestablished.TheresultspresentedinthispapersubstantiallyextendtheresultsduetoChang(2001),Osilikeandhniagbosor(2000),Rhoades(1994)andSchu(1991).

简介：Amorerelaxedsufficientconditionfortheconvergenceoffiltered-XLMS(FXLMS)algorithmispresented.Itispointedoutthatifsomepositiverealconditionforsecondarypathtransferfunctionanditsestimatesissatisfiedwithinallthefrequencybands,FXLMSalgorithmconvergeswhateverthereferencesignalislike.Butiftheabovepositiverealconditionissatisfiedonlywithinsomefrequencybands,theconvergenceofFXLMSalgorithmisdependentonthedistributionofpowerspectraldensityofthereferencesignal,andtheconvergencestepsizeisdeterminedbythedistributionofsomespecificcorrelationmatrixeigenvalues.ApplyingtheconclusionabovetotheDelayedLMS(DLMS)algorithm,itisshownthatDLMSalgorithmwithsomeerroroftimedelayestimationconvergesincertaindiscretefrequencybands,andthewidthofwhicharedeterminedonlybythe"time-delayestimationerrorfrequency"whichisequaltoonefourthoftheinverseofestimatederrorofthetimedelay.

简介：没有标记数据的摆动，错误减小，集中和optimality为散开方程为适应混合有限元素方法(AMFEM)被分析。第一，伪错误，即应力变量误差的和和放大误差评估者，与在二个连续的适应的环之间的一个固定因素被显示出到还原剂，直到摆动。第二，AMFEM的集中加流动错误的分叉关于伪错误被获得。最后，伪最佳的集中率为全部的错误被建立，即加数据摆动的压力变量错误。

简介：InthisarticleweintroducethedifferencesequencespacesW0[f,Δm],W1[f,Δm],W∞[f,Δm]andS[f,Δm],definedbyamodulusfunctionf.WeobtainarelationbetweenW1[f,Δm]∩l∞[f,Δm]andS[f,Δm]∩l∞[f,Δm]andprovesomeinclusionresults.

简介：Inthispaper,basedontheresulspresentedinpartIofthispaper[18],wepresentanumericalcrabedingalgorithmforsolingthenonlinearcomplementarityproblem,andproveitsconvergencecarefully.Numericalexperimentsshowthatthealgorithmissuccessful.

简介：TheInternetevolvestoincorporatevery-high-bandwidthopticallinksandmorelarge-delaysatellitelinks.TCPfacesnewchallengesinthisuniqueenvironment.TheoryandexperimentsshowedthatTCPbecomesinefficientandispronetobeunstableastheper-flowproductofbandwidthandlatencyincreases,regardlessofthequeuingscheme.Variable-structurecongestionControlProtocol(VCP)isproposedtoaddresstheseproblems.However,VCPhasproblemintermsofconvergencetime,i.e.,ittakesalongtimeforanewVCPflowtoachievefairbandwidthallocationiftheexistingVCPflowshavelargecon-gestionwindows.ThispaperproposedanExtendedVariable-structurecongestionControlProtocol(EVCP),whichadoptedaconvergencecontroller.Thebasicideaofconvergencecontrolleristhatifaflowhaslargerwindowthanitsfairwindow,itscongestionwindowshouldbedecreasedmoreaggressivelythanusualinMultiplicativeDecrease(MD)phase.SimulationsshowedthatEVCPhasbetterperformanceintermsofconvergencetimewhilekeepingtheadvantagesofVCP.

简介：ImplicitRunge-Kuttamethodishighlyaccurateandstableforstiffinitialvalueproblem.ButtheiterationtechniqueusedtosolveimplicitRunge-Kuttamethodrequireslotsofcomputationalefforts.Inthispaper,weextendtheParallelDiagonalIteratedRungeKutta(PDIRK)methodstodelaydifferentialequations(DDEs).WegivetheconvergenceregionofPDIRKmethods,andanalyzethespeedofconvergenceinthreepartsfortheP-stabilityregionoftheRunge-Kuttacorrectormethod.Finally,weanalysisthespeed-upfactorthroughanumericalexperiment.TheresultsshowthatthePDIRKmethodstoDDEsareefficient.

简介：TheconvergenceproblemofthefamilyofEuler-HalleymethodsisconsideredundertheLipschitzconditionwiththeL-average,andaunitedconvergencetheorywithitsapplicationsispresented.

简介：调查一个明确的迎风的有限元素方法的集中，平均边的有限元素方法到多维的分级的能量守恒定律。有限元素计划和主要集中定理的讨论；技术词根的证明。

简介：

简介：我们与弱Dirichlet边界条件为时间依赖者Hamilton-Jacobi方程学习数字方法。我们首先建议抽象单调近似计划的一个新班并且得到1/2的集中率。然后，根据摘要，集中结果，由最新构造单调内部和边界点上的有限体积近似，我们与弱Dirichlet边界条件为时间依赖者Hamilton-Jacobi方程获得会聚的有限体积计划。最后给一些数字结果。[从作者抽象]

简介：

简介：

简介：Inthispaper,wefirstprovideageneralizeddifferencemethodforthe2-dimensionalNavier-Stokesequationsbycombingtheideasofstaggeredscheme[6]mandgeneralizedupwindscheme[4]inspace,andbybackwardEulertime-stepping.Thenweapplytheabstractframeworkof[7]toproveitslong-timeconvergence.Finally,anumericalexampleforsolvingdrivencavityflowsisgiven.

简介：Weproposeanovelnumericalapproachfordelaydifferentialequationswithvanishingproportionaldelaysbasedonspectralmethods.ALegendre-collocationmethodisem-ployedtoobtainhighlyaccuratenumericalapproximationstotheexactsolution.Itisprovedtheoreticallyanddemonstratednumericallythattheproposedmethodconvergesexponentiallyprovidedthatthedatainthegivenpantographdelaydifferentialequationaresmooth.

简介：在探索了延期的数字答案的研究上集中微分方程。数字方法的线性稳定性；一个腿方法的申请；错误分析。

简介：这份报纸开发一个新方法在二和三种尺寸在矩形的格子上分析有限元素方法的重申的缺点修正计划的集中。主要想法是提出精力内部的产品和精力(半)标准进矩阵形式。然后，包含的二关键不平等的二个常数是min，二的最大特征值分别地联系概括特征值问题。这二个概括特征值问题的元素水平上的本地版本确切被解决突然地获得(更低)这二个常数的上面的界限。这和为重申的答案的一些必要观察在减少的2D和单调建立集中在3D的性质。为二种尺寸，结果此处在文学改进那些；为三种尺寸，结果此处是新的。数字结果被介绍检验理论结果。[从作者抽象]