简介：Theauthorsderivelawsoftheiteratedlogarithmforkernelestimatorofregressionfunctionbasedondirectionaldata.Theresultsaredistributionfreeinthesensethattheyaretrueforalldistributionsofdesignvariable.

简介：我们认为在线性、非线性的平常的微分方程(颂诗)的参数评价的问题当模特儿。非线性的颂诗模型广泛地在应用被使用。但是他们的分析答案通常不是可得到的。因此常规的方法通常取决于带巨大的计算费用的数字答案的重复使用。我们建议了在第二个阶段在第一个阶段，和一个数字discretization方法(Eulersdiscretization方法，trapezoidaldiscretization方法，或Runge-Kuttadiscretization方法)包括一个变光滑的方法(变光滑的核或本地多项式试穿)的一条新二阶段的途径。通过数字模拟，我们发现建议方法获得在评价精确性和计算费用之间的合适的平衡。Asymptotic性质也被介绍，它在一些温和条件下面显示出一致性和评估者的asymptotic规度。建议方法以精确性和计算费用与存在方法相比。模拟结果证明有在第二个阶段的在第一个阶段和trapezoidaldiscretization的本地线性变光滑的评估者有最低平均相对错误。我们把建议方法用于HIV动力学数据说明评估者的有实行可能。

简介：Segmentationofthree-dimensional(3D)complicatedstructuresisofgreatimportanceformanyrealapplications.Inthisworkwecombinegraphcutminimizationmethodwithavariantofthelevelsetideafor3DsegmentationbasedontheMumford-Shahmodel.ComparedwiththetraditionalapproachforsolvingtheEuler-Lagrangeequationwedonotneedtosolveanypartialdifferentialequations.Instead,theminimumcutonaspecialdesignedgraphneedtobecomputed.Themethodistestedondatawithcomplicatedstructures.Itisratherstablewithrespecttoinitialvalueandthealgorithmisnearlyparameterfree.Experimentsshowthatitcansolvelargeproblemsmuchfasterthantraditionalapproaches.

简介：Inthispaperweconsideranon-standardinverseheatconductionproblemfordeterminingsurfaceheatfluxfromaninteriorobservationwhichappearsinsomeappliedsubjects.Thisproblemisill-posedinthesensethatthesolution(ifitexists)doesnotdependcontinuouslyonthedata.AFouriermethodisappliedtoformulatearegularizedapproximationsolution,andsomesharperrorestimatesarealsogiven.

简介：这篇文章学习最佳的消费空闲，公事包和其偏爱被-maxmin提出的一个无穷地过的投资者的退休选择是区分歧义和歧义态度的期望的CES用途。采用递归的multiplepriors用途和向后的随机的微分方程(BSDE)的技术，我们转变-maxmin期望的CES用途进在与投资者无常的度有关的一项新概率措施下面的一个古典期望的CES实用程序。我们的模型调查最佳的consumption-leisure-work选择，最佳的公事包选择，和最佳的停止问题。在这个模型，投资者能与一种退休选择一起在某个最小的工作小时上面灵活地调整她劳动的供应。问题能是使用变化不平等解决的经分解。并且当她的财富超过某个批评水平时，最佳的退休时间作为第一次被给。在退休前后的最佳的消费空闲和公事包策略在关上的形式被提供。最后，在没有暧昧的从那些的歧义下面的最佳的消费空闲，公事包和批评财富水平的区别被讨论。

简介：

简介：Inthispaper,anewdomaindecompositionmethodbasedonthenaturalboundaryreduction,whichsolveswaveproblemsoveranunboundeddomain,issuggestted.Ancircularartifcialboundaryisintroduced.Theoriginalunboundeddomainisdividedintotwosubdomains,aninternalboundedregionandexternalunboundedregionoutsidetheartificialboundary.ADirichlet-Neumann(D-N)alternatingiterationalgorithmisconstructed.WeprovethatthealgorithmisequavilenttopreconditionalRichardsoniterationmethod.Numericalstudiesareperformedbyfiniteelementmethod.ThenumericalresultsshowthattheconvergencerateofthediscreteD-Niterationisindependentofthefniteelementmeshsize.

简介：ForthenumericalevaluationofCauchyprincipalvalueintegralsoftheformf(x)(x-λ)-1dxwithλ∈(-1,1)andf∈C1[-1,1],weinvestigatethequadratureformulaQ(n+1)Spl1[·;λ]obtainedbyreplacingtheintegrandfunctionfbyitspiecewiselinearinterpolantatanequidistantsetofnodesasproposedbyRabinoivitz(Math.Comp.,51:741-747,1988).WegiveupperboundsforthePeano-typeerrorconstantsfors∈{1,2}.ThesearethebestpossibleconstantsininequalitiesofthetypeFurthermore,weprovethatourupperboundsareasymptoticallysharp.

简介：Inthisworkweconsidertheproblemofshapereconstructionfromanunorganizeddatasetwhichhasmanyimportantapplicationsinmedicalimaging,scientificcomputing,reverseengineeringandgeometricmodelling.ThereconstructedsurfaceisobtainedbycontinuouslydeforminganinitialsurfacefollowingthePartialDifferentialEquation(PDE)-baseddiffusionmodelderivedbyaminimalvolume-likevariationalformulation.Theevolutionisdrivenbothbythedistancefromthedatasetandbythecurvatureanalyticallycomputedbyit.Thedistancefunctioniscomputedbyimplicitlocalinterpolantsdefinedintermsofradialbasisfunctions.SpacediscretizationofthePDEmodelisobtainedbyfiniteco-volumeschemesandsemi-implicitapproachisusedintime/scale.Theuseofalevelsetmethodforthenumericalcomputationofthesurfacereconstructionallowsustohandlecomplexgeometryandevenchangingtopology,withouttheneedofuser-interaction.Numericalexamplesdemonstratetheabilityoftheproposedmethodtoproducehighqualityreconstructions.Moreover,weshowtheeffectivenessofthenewapproachtosolveholefillingproblemsandBooleanoperationsbetweendifferentdatasets.

简介：Klapper(1994)showedthatthereexistsaclassofgeometricsequenceswiththemaximalpossiblelinearcomplexitywhenconsideredassequencesoverGF(2),butthesesequenceshaveverylowlinearcomplexitieswhenconsideredassequencesoverGF(p)(pisanoddprime).ThislinearcomplexityofabinarysequencewhenconsideredasasequenceoverGF(p)iscalledGF(p)complexity.ThisindicatesthatthebinarysequenceswithhighGF(2)linearcomplexitiesareinadequateforsecurityinthepracticalapplication,while,theirGF(p)linearcomplexitiesarealsoequallyimportant,evenwhentheonlyconcerniswithattacksusingtheBerlekamp-Masseyalgorithm[Massey,J.L.,Shift-registersynthesisandbchdecoding,IEEETransactionsonInformationTheory,15(1),1969,122–127].Fromthisperspective,inthispapertheauthorsstudytheGF(p)linearcomplexityofHall’ssexticresiduesequencesandsomeknowncyclotomic-set-basedsequences.

简介：Inthispaper,themodifiedextendedtanhmethodisusedtoconstructmoregeneralexactsolutionsofa(2+1)-dimensionalnonlinearSchr¨odingerequation.WiththeaidofMapleandMatlabsoftware,weobtainexactexplicitkinkwavesolutions,peakonwavesolutions,periodicwavesolutionsandtheir3Dimages.

简介：Inthispapertheauthorpresentsamethodforthenumericalsolutionofa2-DCauchyprincipalvalueoftheformwhereSisadomainwithacontinuousboundary.Byusmgpolarcoordinates,theintegralisreducedtotheformwheredenotesthefinite-partoftheintegral.Weconstructtherelativeproductrulebasedonquasi-interpolatingsplines.Convergenceresultsareprovedandnumericalexamplesaregiven.

简介：有牛顿重复的一个新decoupled二系结算法为解决描述通过多孔的媒介过滤的液体流动的联合Navier-Stokes/Darcy模型被建议。而且错误估计被给，当时，它证明精确性的一样的顺序能作为在好网孔直接解决系统被完成h=H2。理论分析和数字实验为解决联合问题说明算法的效率。

简介：

简介：Anonlinearparabolicsystemisderivedtodescribeincompressiblenuclearwaste-disposalcontaminationinporousmedia.Asequentialimplicittirne-steppingisdefined,inwhichthepressureandDarcyvelocityofthemixtureareapproximatedsimultaneouslybyamixedfiniteelementmethodandthebrine,radionuclidandheataretreatedbyacombinationofaGalerkinfiniteelementmethodandthemethodofcharacteristics.Optimal-orderconvergenceinL2isproved.Time-truncationerrorsofstandardproceduresarereducedbytimesteppingalongthecharacteristicsofthehyperbolicpartofthebrine,radionuclideandhealequalios,temporalandspatialerrorarelossenedbydirectcompulationofthevelocityinthemixedmethod,asopposedtodifferentiationofthepressure.

简介：

简介：一个数字算法为包含与排干和涂层流动联系的第三顺序的平常的微分方程的某些边界价值问题的答案的近似被开发。数字算法获得的近似解决方案用nonpolynomialquintic花键函数发展了由的作者表演比另外的花键和域分解方法生产的那些好。算法在与排干和涂层流动表明途径的实际实用性联系的二个问题上被测试。

简介：

简介：Ahigh-orderleap-frogbasednon-dissipativediscontinuousGalerkintime-domainmethodforsolvingMaxwell'sequationsisintroducedandanalyzed.Theproposedmethodcombinesacenteredapproximationfortheevaluationoffluxesattheinterfacebetweenneighboringelements,withaNth-orderleap-frogtimescheme.Moreover,theinterpolationdegreeisdefinedattheelementlevelandthemeshisrefinedlocallyinanon-conformingwayresultinginarbitrarylevelhangingnodes.ThemethodisprovedtobestableundersomeCFL-likeconditiononthetimestep.Theconvergenceofthesemi-discreteapproximationtoMaxwell'sequationsisestablishedrigorouslyandboundsontheglobaldivergenceerrorareprovided.Numericalexperimentswithhigh-orderelementsshowthepotentialofthemethod.