简介:
简介:LetHbetherealquaternionfield,CandRbethecomplexandrealfieldrespectively.ClearlyR(?)C(?)H.LetHm×ndenotethesetofallm×nmatricesoverH.IfA=(ars)∈Hm×n,thenthereexistA1andA2∈Cm×nsuchthatA=A1+A2j.LetACdenotethecomplexrepresentationofA,thatisthe2m×2ncomplexmatrixAc=((A1/A2)(-A2/A1))(see[1,2]).WedenotebyADtheDrazininverseofA∈Hm×nwhichistheuniquesolutionofthee-
简介:ByusingMoore-Penrosegeneralizedinverseandthegeneralsingularvaluedecompositionofmatrices,thispaperestablishesthenecessaryandsufficientconditionsfortheexistenceofandtheexpressionsforthecentrosymmetricsolutionswithasubmatrixconstraintofmatrixinverseproblemAX=B.Inaddition,inthesolutionsetofcorrespondingproblem,theexpressionoftheoptimalapproximationsolutiontoagivenmatrixisderived.
简介:isgainedbydeletingthekthrowandthekthcolumn(k=1,2,...,n)fromTn.Weputfor-wardaninverseeigenvalueproblemtobethat:Ifwedon’tknowthematrixT1,n,butweknowalleigenvaluesofmatrixT1,k-1,alleigenvaluesofmatrixTk+1,k,andalleigenvaluesofmatrixT1,ncouldweconstructthematrixT1,n.Letμ1,μ2,…,μk-1,μk,μk+1,…,μn-1,
简介:Theinversescatteringproblemsaretodetectthepropertyofobstaclesfromthemeasurementsoutsidetheobstacles.Oneofimportantresearchareasinthistopicistherecoveryofboundarypropertyforimpenetrableobstacles.Inthispaper,wewouldliketogiveabriefreviewabouttherecentlydevelopedsingularsourcemethods.Therearethreedifferentmethodsinthiscategory,namely,linearsamplingmethod,pointsourcemethodandprobemethod.Wealsopresentsomerecentnewresultsabouttheprobemethod.
简介:Acompoundneuralnetworkisutilizedtoidentifythedynamicnonlinearsystem.Thisnetworkiscomposedoftwoparts:oneisalinearneuralnetwork,andtheotherisarecurrentneuralnetwork.Basedontheinversetheoryacompoundinversecontrolmethodisproposed.Thecontrollerhasalsotwoparts:alinearcontrollerandanonlinearneuralnetworkcontroller.Thestabilityconditionoftheclosed-loopneuralnetwork-basedcompoundinversecontrolsystemisdemonstratedbasedontheLyapunovtheory.Simulationstudieshaveshownthatthisschemeissimpleandhasgoodcontrolaccuracyandrobustness.
简介:
简介:Inversetechniqueisawidelyusedmethodinoceanography,butithasaproblemthattheretrievedsolutionsoftenviolatemodelpriorassumptions.Totunethemodelhasconsistentsolutions,aniterationapproach,whichsuccessivelyutilizestheposteriorstatisticsfornextroundinverseestimation,isintroducedandtestedfromarealcasestudy.Itisfoundthattheconsistencymaybecomeelusiveasthedeterminantsofsolutionandnoisecovariancematricesbecomezerointheiterationprocess.However,afterseveralstepsofsuchoperation,thedifferencebetweenposteriorstatisticsandthemodelprioronescanbegraduallyreduced.
简介:AnewmethodforapproximatingtheinerseLaplacetransformispresented.WefirstchangeourLaplacetransformequationintoaconvolutiontypeintegralequation,whereTikhonovregularizationtechniquesandtheFouriertransformationareeasilyapplied.WefinallyobtainaregularizedapproximationtotheinverseLaplacetransformasfinitesum
简介:TheBi-LSmethodbasedonQRdecompositionprovidesaconvenientframeworkforde-velopingefficientsubspacetrackingalgorithms.ToovercometheshortcomingofthebacksubstitutionstepandimprovetheparallelarchitectureoftheBi-LSalgorithms,aBi-LSsubspacetrackingalgorithmbasedonInverseQR(IQR)decompositionisdeveloped.TheproposedIQRiterativealgorithmforsubspacetrackingiswellsuitedfortheparallelimplementationinthesystolicarray.SimulationresultsarepresentedtoillustratetheeffectivenessoftheproposedIQRsubspacetrackingalgorithm.