/
NAGLibraryRoutineDocumentF08YEFDTGSJAbeforeusingthisroutinepleasereadt NAGLibraryRoutineDocumentF08YEFDTGSJAbeforeusingthisroutinepleasereadt

NAGLibraryRoutineDocumentF08YEFDTGSJAbeforeusingthisroutinepleasereadt - PDF document

blanko
blanko . @blanko
Follow
342 views
Uploaded On 2021-08-05

NAGLibraryRoutineDocumentF08YEFDTGSJAbeforeusingthisroutinepleasereadt - PPT Presentation

Optionallymayormaynotbecomputedortheymaybepremultipliedbymatricesrespectively0thenhavetheform0thenhavetheform000Inbothcasesthediagonalmatrixhasnonnegativediagonalelementsthediagonalmatrixpositivediag ID: 857572

nag nout write kind nout nag kind write real integer inputonentry dtgsja beta lda alpha ifail arrayinput max iscalled

Share:

Link:

Embed:

Download Presentation from below link

Download Pdf The PPT/PDF document "NAGLibraryRoutineDocumentF08YEFDTGSJAbef..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.


Presentation Transcript

1 NAGLibraryRoutineDocumentF08YEF(DTGSJA)b
NAGLibraryRoutineDocumentF08YEF(DTGSJA)beforeusingthisroutine,pleasereadtheUsers'Noteforyourimplementationtochecktheinterpretationofbolditalicisedtermsandotherimplementation-dependentdetails.1Purpose Optionallymayormaynotbecomputed,ortheymaybepremultipliedbymatricesrespectively.ðÞ0thenhavetheform;;;...;;;...;;;0thenhavetheform000;;;...;;;...;;;Inbothcasesthediagonalmatrixhasnon-negativediagonalelements,thediagonalmatrixpositivediagonalelements,sothatisnonsingular,and1.SeeSection2.3.5.3ofAndersonetal.(1999)forfurtherinformation.4ReferencesAndersonE,BaiZ,BischofC,BlackfordS,DemmelJ,DongarraJJ,DuCrozJJ,GreenbaumA,HammarlingS,McKenneyAandSorensenD(1999)LAPACKUsers'Guide(3rdEdition)SIAM,Philadelphiahttp://www.netlib.org/lapack/lugGolubGHandVanLoanCF(1996)MatrixComputations(3rdEdition)JohnsHopkinsUniversityPress,Baltimore5ArgumentsInputOnentry:ifJOBUU,Umustcontainanorthogonalmatrixonentry,andtheproductisreturned.IfJOBUI,Uisinitializedtotheunitmatrix,andtheorthogonalmatrixisreturned.IfJOBUisnotcomputed.:JOBUU,IorN. NA

2 GLibraryManualF08YEF.2Mark26 InputOnentr
GLibraryManualF08YEF.2Mark26 InputOnentry:ifJOBVV,Vmustcontainanorthogonalmatrixonentry,andtheproductisreturned.IfJOBVI,Visinitializedtotheunitmatrix,andtheorthogonalmatrixisreturned.IfJOBVisnotcomputed.:JOBVV,IorN.InputOnentry:ifJOBQQ,Qmustcontainanorthogonalmatrixonentry,andtheproductisreturned.IfJOBQI,Qisinitializedtotheunitmatrix,andtheorthogonalmatrixisreturned.IfJOBQisnotcomputed.:JOBQQ,IorN.InputOnentry,thenumberofrowsofthematrixINTEGERInputOnentry,thenumberofrowsofthematrixINTEGERInputOnentry,thenumberofcolumnsofthematricesInputINTEGERInputOnentry:KandLspecifythesizes,,ofthesubblocksof,whoseGSVDistobecomputedbyF08YEF(DTGSJA).REAL(KIND=nag_wp)arrayInput/Output:theseconddimensionofthearrayAmustbeatleastmax1Onentry:thematrixOnexit:if0,Al;ncontainstheuppertriangularmatrix0,Am;ncontainstherowsoftheuppertriangularmatrix,andthesubmatrixisreturnedinl;n10:InputOnentry:therstdimensionofthearrayAasdeclaredinthe(sub)programfromwhichF08YEF(DTGSJA)iscalled.:LDAmax1 Least-squaresandEigenvalueProblems(LAPACK)Mark26F08YEF.3

3 11:REAL(KIND=nag_wp)arrayInput/Output:th
11:REAL(KIND=nag_wp)arrayInput/Output:theseconddimensionofthearrayBmustbeatleastmax1Onentry:theOnexit:if0,Bl;ncontainsthesubmatrix12:InputOnentry:therstdimensionofthearrayBasdeclaredinthe(sub)programfromwhichF08YEF(DTGSJA)iscalled.:LDBmax113:TOLAREAL(KIND=nag_wp)Input14:TOLBREAL(KIND=nag_wp)InputOnentry:TOLAandTOLBaretheconvergencecriteriafortheJacobiKogbetliantziterationprocedure.Generally,theyshouldbethesameasusedinthepreprocessingstepperformedbyF08VEF(DGGSVP)orF08VGF(DGGSVP3),sayTOLAmaxMmaxPisthemachineprecision15:REAL(KIND=nag_wp)arrayOutputOnexit:seethedescriptionofBETA.16:REAL(KIND=nag_wp)arrayOutputOnexit:ALPHAandBETAcontainthegeneralizedsingularvaluepairsof1,BETA0,for...,and0,ALPHA,BETA,for...,or0,ALPHA,BETA,for...0,BETA1,for...Furthermore,if,ALPHA0,for...17:REAL(KIND=nag_wp)arrayInput/Output:theseconddimensionofthearrayUmustbeatleastmax1ifJOBUUorI,andatleast1otherwise.Onentry:ifJOBUU,Umustcontainan(usuallytheorthogonalmatrixreturnedbyF08VEF(DGGSVP)orF08VGF(DGGSVP3)).Onexit:ifJOBUU,UcontainstheproductIfJOBUI

4 ,UcontainstheorthogonalmatrixIfJOBUN,Uis
,UcontainstheorthogonalmatrixIfJOBUN,Uisnotreferenced.18:InputOnentry:therstdimensionofthearrayUasdeclaredinthe(sub)programfromwhichF08YEF(DTGSJA)iscalled.ifJOBUUorI,LDUmax1otherwiseLDU NAGLibraryManualF08YEF.4Mark26 19:REAL(KIND=nag_wp)arrayInput/Output:theseconddimensionofthearrayVmustbeatleastmax1ifJOBVVorI,andatleast1otherwise.Onentry:ifJOBVV,Vmustcontainan(usuallytheorthogonalmatrixreturnedbyF08VEF(DGGSVP)orF08VGF(DGGSVP3)).Onexit:ifJOBVI,VcontainstheorthogonalmatrixIfJOBVV,VcontainstheproductIfJOBVN,Visnotreferenced.20:InputOnentry:therstdimensionofthearrayVasdeclaredinthe(sub)programfromwhichF08YEF(DTGSJA)iscalled.ifJOBVVorI,LDVmax1otherwiseLDV21:REAL(KIND=nag_wp)arrayInput/Output:theseconddimensionofthearrayQmustbeatleastmax1ifJOBQQorI,andatleast1otherwise.Onentry:ifJOBQQ,Qmustcontainan(usuallytheorthogonalmatrixreturnedbyF08VEF(DGGSVP)orF08VGF(DGGSVP3)).Onexit:ifJOBQI,QcontainstheorthogonalmatrixIfJOBQQ,QcontainstheproductIfJOBQN,Qisnotreferenced.22:InputOnentry:therstdimensionofthearrayQasdeclaredinthe(sub

5 )programfromwhichF08YEF(DTGSJA)iscalled.
)programfromwhichF08YEF(DTGSJA)iscalled.ifJOBQQorI,LDQmax1otherwiseLDQ23:REAL(KIND=nag_wp)arrayWorkspace24:INTEGEROutputOnexit:thenumberofcyclesrequiredforconvergence.25:OutputOnexit:INFO0unlesstheroutinedetectsanerror(seeSection6).6ErrorIndicatorsandWarningsIfINFO,argumenthadanillegalvalue.Anexplanatorymessageisoutput,andexecutionoftheprogramisterminated. Least-squaresandEigenvalueProblems(LAPACK)Mark26F08YEF.5 Theproceduredoesnotconvergeafter40cycles.7AccuracyThecomputedgeneralizedsingularvaluedecompositionisnearlytheexactgeneralizedsingularvaluedecompositionfornearbymatrices,whereandisthemachineprecision.SeeSection4.12ofAndersonetal.(1999)forfurtherdetails.8ParallelismandPerformanceF08YEF(DTGSJA)makescallstoBLASand/orLAPACKroutines,whichmaybethreadedwithinthevendorlibraryusedbythisimplementation.ConsultthedocumentationforthevendorlibraryforfurtherPleaseconsulttheX06ChapterIntroductionforinformationonhowtocontrolandinterrogatetheOpenMPenvironmentusedwithinthisroutine.PleasealsoconsulttheUsers'Noteforyourimplementa

6 tionforanyadditionalimplementation-speci
tionforanyadditionalimplementation-specicinformation.9FurtherCommentsThecomplexanalogueofthisroutineisF08YSF(ZTGSJA).10ExampleThisexamplendsthegeneralizedsingularvaluedecompositionofthematrixpairA;B,where12332145678846510.1ProgramTextProgramf08yefe!F08YEFExampleProgramText!Mark26Release.NAGCopyright2016.!..UseStatements..Usenag_library,Only:dggsvp,dtgsja,f06raf,nag_wp,x02ajf,x04cbf!..ImplicitNoneStatement..ImplicitNone!..Parameters..Integer,Parameter::nin=5,nout=6!..LocalScalars..Real(Kind=nag_wp)::eps,tola,tolbInteger::i,ifail,info,irank,j,k,l,lda,&ldb,ldq,ldu,ldv,m,n,ncycle,p!..LocalArrays..Real(Kind=nag_wp),Allocatable::a(:,:),alpha(:),b(:,:),beta(:),&q(:,:),tau(:),u(:,:),v(:,:),&Integer,Allocatable::iwork(:)Character(1)::clabs(1),rlabs(1)!..IntrinsicProcedures..NAGLibraryManualF08YEF.6Mark26 Intrinsic::max,real!..ExecutableStatements..Write(nout,*)’F08YEFExampleProgramResults’Write(nout,*)Flush(nout)!SkipheadingindatafileRead(nin,*)Read(nin,*)m,n,plda=mldb=pldq=nldu=mldv=pAllocate(a(lda,n),alpha(n),b(ldb,n),beta

7 (n),q(ldq,n),tau(n),u(ldu,m),&v(ldv,p),w
(n),q(ldq,n),tau(n),u(ldu,m),&v(ldv,p),work(m+3*n+p),iwork(n))!ReadthembynmatrixAandpbynmatrixBfromdatafileRead(nin,*)(a(i,1:n),i=1,m)Read(nin,*)(b(i,1:n),i=1,p)!Computetolaandtolbas!tola=max(m,n)*norm(A)*macheps!tolb=max(p,n)*norm(B)*machepseps=x02ajf()tola=real(max(m,n),kind=nag_wp)*f06raf(’One-norm’,m,n,a,lda,work)*epstolb=real(max(p,n),kind=nag_wp)*f06raf(’One-norm’,p,n,b,ldb,work)*eps!Computethefactorizationof(A,B)!(A=U1*S*(Q1**T),B=V1*T*(Q1**T))!TheNAGnameequivalentofdggsvpisf08vefCalldggsvp(’U’,’V’,’Q’,m,p,n,a,lda,b,ldb,tola,tolb,k,l,u,ldu,v,ldv,q,&ldq,iwork,tau,work,info)!Computethegeneralizedsingularvaluedecompositionof(A,B)!(A=U*D1*(0R)*(Q**T),B=V*D2*(0R)*(Q**T))!TheNAGnameequivalentofdtgsjaisf08yefCalldtgsja(’U’,’V’,’Q’,m,p,n,k,l,a,lda,b,ldb,tola,tolb,alpha,beta,u,&ldu,v,ldv,q,ldq,work,ncycle,info)If(info==0)Then!PrintsolutionWrite(nout,*)’Numberofinfinitegeneralizedsingularvalues(K)’Write(nout,99999)kWrite(nout,*)’Numberoffinitegeneralizedsingularvalues(L)’Write(nout,99999)lWrite(nout,*)’EffectiveNumeric

8 alrankof(A**TB**T)**T(K+L)’Write(nout,99
alrankof(A**TB**T)**T(K+L)’Write(nout,99999)irankWrite(nout,*)Write(nout,*)’Finitegeneralizedsingularvalues’Write(nout,99998)(alpha(j)/beta(j),j=k+1,irank)Write(nout,*)Flush(nout)!ifail:behaviouronerrorexit!=0forhardexit,=1forquiet-soft,=-1fornoisy-softifail=0Callx04cbf(’General’,’’,m,m,u,ldu,’1P,E12.4’,’OrthogonalmatrixU’,&’Integer’,rlabs,’Integer’,clabs,80,0,ifail)Write(nout,*)Flush(nout)Callx04cbf(’General’,’’,p,p,v,ldv,’1P,E12.4’,’OrthogonalmatrixV’,&’Integer’,rlabs,’Integer’,clabs,80,0,ifail)Least-squaresandEigenvalueProblems(LAPACK)Mark26F08YEF.7 Write(nout,*)Flush(nout)Callx04cbf(’General’,’’,n,n,q,ldq,’1P,E12.4’,’OrthogonalmatrixQ’,&’Integer’,rlabs,’Integer’,clabs,80,0,ifail)Write(nout,*)Flush(nout)Callx04cbf(’Uppertriangular’,’Non-unit’,irank,irank,a(1,n-irank+1),&lda,’1P,E12.4’,’NonsingularuppertriangularmatrixR’,’Integer’,&rlabs,’Integer’,clabs,80,0,ifail)Write(nout,*)Write(nout,*)’NumberofcyclesoftheKogbetliantzmethod’Write(nout,99999)ncycleWrite(nout,99997)’FailureinDTGSJA.INFO=’,infoEndIf99999Format(1X

9 ,I5)99998Format(3X,8(1P,E12.4))99997Form
,I5)99998Format(3X,8(1P,E12.4))99997Format(1X,A,I4)EndProgramf08yefe10.2ProgramDataF08YEFExampleProgramData432:ValuesofM,NandP1.02.03.03.02.01.04.05.06.07.08.08.0:EndofmatrixA-2.0-3.03.04.06.05.0:EndofmatrixB10.3ProgramResultsF08YEFExampleProgramResultsNumberofinfinitegeneralizedsingularvalues(K)Numberoffinitegeneralizedsingularvalues(L)EffectiveNumericalrankof(A**TB**T)**T(K+L)Finitegeneralizedsingularvalues1.3151E+008.0185E-02OrthogonalmatrixU12341-1.3484E-015.2524E-01-2.0924E-018.1373E-0126.7420E-01-5.2213E-01-3.8886E-013.4874E-0132.6968E-015.2757E-01-6.5782E-01-4.6499E-0146.7420E-014.1615E-016.1014E-011.5127E-15OrthogonalmatrixV13.5539E-01-9.3472E-0129.3472E-013.5539E-01OrthogonalmatrixQ1-8.3205E-01-9.4633E-02-5.4657E-0125.5470E-01-1.4195E-01-8.1985E-0130.0000E+00-9.8534E-011.7060E-01NAGLibraryManualF08YEF.8Mark26 NonsingularuppertriangularmatrixR1-2.0569E+00-9.0121E+00-9.3705E+002-1.0882E+01-7.2688E+003-6.0405E+00NumberofcyclesoftheKogbetliantzmethodLeast-squaresandEigenvalueProblems(LAPACK)Mark26F08YEF.9(last)

Related Contents


Next Show more