简介：

简介：

简介：在时间域的领域分解和匹配的方法为模仿在波浪进展的轮船的运动被构画出。流动域被想象的控制表面分解成内部、外部的域，并且当短暂格林函数方法在外部域被使用时，Rankine来源方法被用于内部域。二个起始的边界价值问题在控制表面上被匹配。相应数字代码被开发，并且为系列60轮船和S175货柜船在头海进展的增加的群众，波浪激动力量和轮船运动，被介绍并且验证。当数字结果与试验性的数据和另外的引用相比时，一个好协议被获得了。它证明因为面板discretization，现在的方法在数字计算期间在内部领域仅仅是更有效的，并且好数字稳定性被证明与闪光关于轮船避免分叉问题。

简介：SeveralcubatureformulasonthecubicdomainsarederivedusingthediscreteFourieranalysisassociatedwithlatticetiling,asdevelopedin[10].ThemainresultsconsistofanewderivationoftheGaussiantypecubaturefortheproductChebyshevweightfunctionsandassociatedinterpolationpolynomialson[-1,1]~2,aswellasnewresultson[-1,1]~3.Inparticular,compactformulasforthefundamentalinterpolationpolynomialsarederived,basedonn~3/4+(?)(n~2)nodesofacubatureformulaon[-1,1]~3.

简介：

简介：Thissurveyreviewstherecentdevelopmentofgradientdomainmeshdeformationmethod.Differenttootherdeformationmethods,thegradientdomaindeformationmethodisasurface-based,variationaloptimizationmethod.Itdirectlyencodesthegeometricdetailsindifferentialcoordinates,whicharealsocalledLaplaciancoordinatesinliterature.BypreservingtheLaplaciancoordinates,themeshdetailscanbewellpreservedduringdeformation.DuetothelocalityoftheLaplaciancoordinates,thevariationaloptimizationproblemcanbecastedintoasparselinearsystem.Fastsparselinearsolvercanbeadoptedtogeneratedeformationresultinteractively,oreveninreal-time.Thenonlinearnatureofgradientdomainmeshdeformationleadstothedevelopmentoftwocategoriesofdeformationmethods:linearizationmethodsandnonlinearoptimizationmethods.Basically,thelinearizationmethodsonlyneedtosolvethelinearleast-squaressystemonce.Theyarefast,easytounderstandandcontrol,whilethedeformationresultmightbesuboptimal.Nonlinearoptimizationmethodscanreachoptimalsolutionofdeformationenergyfunctionbyiterativeupdating.Sincethecomputationofnonlinearmethodsisexpensive,reduceddeformablemodelsshouldbeadoptedtoachieveinteractiveperformance.Thenonlinearoptimizationmethodsavoidtheuserburdentoinputtransformationatdeformationhandles,andtheycanbeextendedtoincorporatevariousnonlinearconstraints,likevolumeconstraint,skeletonconstraint,andsoon.Wereviewrepresentativemethodsandrelatedapproachesofeachcategorycomparativelyandhopetohelptheuserunderstandthemotivationbehindthealgorithms.Finally,wediscusstherelationbetweenphysicalsimulationandgradientdomainmeshdeformationtorevealwhyitcanachievephysicallyplausibledeformationresult.

简介：Toacceleratethetrainingofsupportvectordomaindescription(SVDD),confidencesupportvectordomaindescription(CSVDD)isproposedbasedontheobservationthatthedescriptionboundaryisdeterminedbyasmallsubsetoftrainingdatacalledsupportvectors.Namely,thenumberoftrainingsamplesintheuserdefinedsphereiscalculatedandtakenastheconfidencemeasure,accordingtowhichthetrainingsamplesarerankedinascendingorder.ThoseformerrankedonesareselectedastheboundarytargetsfortheSVDDtraining.SimulationsonUCIdatademonstratetheeffectivenessandsuperiorityofCSVDD:thenumberoftrainingtargetsandthetrainingtimearereducedwithoutanylossofaccuracy.

简介：

简介：Anewblock-basedfractalimagecodingalgorithmcalledFractalBlockCodinginResidueDomain(FBCRD)isproposed.InbasicFractalBlockCoding(FBC)algorithm,eachblock(calledrangeblock)isencodedbyanaffinemappingfromadomainblockwithinthesameimagetoitself.Thedecoderusestheparametersofthesemappingstosynthesizethereconstructedimagethroughaniterativeprocedure.FBCRDisamodificationofbasicFBC.InFBCRD,rangeblocksanddomainblocksareallresidueblockssubtractedfromtheirblockmeansandboththeparametersofaffinemappingsandblockmeansarecoded.Thismodificationleadstofeweriterationsatthedecoder.AnoptimizeddecodingstrategyisalsointroducedwhichreducestotaldecodingtimebymorethanhalfofthatofbasicFBC.Thisimprovementisfavorableforrealtimeimplementationoffractalimagecompression.

简介：

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

简介：

简介：Aclassofnonconformingfiniteelementsisconsideredinthispaper,whichiscontinuousonlyatthenodesofthequasi-uniformmesh.Weshowthatthereex-istsanessentialestimatewhichindicatestheequivalencerelation,independentofthemeshparameter,betweentheenergiesofthenonconformingdiscreteharmonicextensionsindifferentsubdomains.Theessentialestimateisofgreatimportanceintheanalysisofthenonoverlappingdomaindecompositionmethodsappliedtosecondorderpartialdifferentialequationsdiscretizedbynonconformingfiniteele-ments.

简介：

简介：空中地磁学的坡度经常被用来提高细节或为解释增加新卓见。坡度可以被测量或源于全部的领域或从在水平、垂直的坡度之间的转变。目前，垂直，水平，并且triaxial空中地磁学的gradiometers在全世界生效，当时开始的，二更广泛地被使用。在水平、垂直的坡度之间的转变为获得三个坡度部件或为检查测量坡度的有效性被需要。由在频率领域的快Fourier变换技术的潜在的地的转变流行地被使用；然而，当适用于在坡度之间转变时，有一个问题，需要解决。因为当u或v等于零或u和v是时，变换操作符的那些表达式是未定义的，同时等于到零(u是在x方向的频率，并且v是在y方向的频率)，操作符不能在这些频率被取样。因而，转变不能被快Fourier变换技术直接实现。在这篇文章，移动采样理论为解决这个问题被采用。模型测试结果证明这种技术有好精确性，并且转变的真实盒子显示计算结果与测量坡度更好同意；它表明方法的有效性而且测量坡度的可靠性不仅。

简介：Singularlyperturbedproblemsforthenonlineardliptiesystemsinastripdomainareconsidered.Undersuitablaconditions,byusingthecomparisontheorem,theexistenceandasymptoticbehaviorofsolutionfortheboundaryvalueproblemsarestudied.

简介：Theslowconvergencerateofreinforcementlearningalgorithmslimitstheirwiderapplication.Inengineeringdomains,hierarchicalreinforcementlearningisdevelopedtoperformactionstemporallyaccordingtopriorknowledge.Thissystemcanconvergefastduetoreducedstatespace.Thereisatestofelevatorgroupcontroltoshowthepowerofthenewsystem.Twoconventionalgroupcontrolalgorithmsareadoptedaspriorknowledge.Performanceindicatesthathierarchicalreinforcementlearningcanreducethelearningtimedramatically.

简介：Domainswitchingisthemainsourceofnonlinearcharacteristicsinferroelectrics.Accordingtocrystalplasticitytheory,thedomainsanddomainswitchingsystemsforperovskitetypestructureferroelectricsaredefined.Consideringthetraversemotionperformanceofdomainwall,arathersimplifiedformofevolutionlawaboutincrementalofvolumefractionduringdomainswitchinghasbeendeveloped.Themainfactors,whichexertaninfluenceondomainswitching,suchasmaterialparameters,domainwallmotionhistory,kindofdomainswitching(180°or90°)andvolumefraction,couldbeaddressed.Thehysteresisloopsofspontaneouselectricpolarizationasafunctionofelectricfield,thebutterflyshapedstrainversuselectricfieldcurveandtheplatformrelationsbetweenspontaneouspolarizationandstress,aswellasthelongitudinalstrainandstress,arewellsimulatedanddiscussed.

简介：Atrecent,Hourgatetgaveadomaindecompositionalgorithmforellipticproblemswhichcanbeimplementedinparallel.Manynumericalexperimentshaveillustrateditsefficiency.Inthepresentpaper,weapplythisalgorithmtosolvethediscreteparabolicproblems,analyseitsconvergenceandshowthatitsconvergenceraleisabout(1-2p+σp2)whichisnearlyoptimalandindependentoftheparameterτ,whereστO((1+H)(1+ln(H/h))2).0

简介：