tensor toolbox手冊(cè)

1、TensToolboxfdensesparsedecomposednwayarrays.cp_alsComputeaCPdecompositionofanytypeoftens.ALS交替最小二乘法求張量CP分解P=CP_ALS(XR)——計(jì)算張量X秩為R的最佳近似CP分解，P=[P.lambdaP.U]P=CP_ALS(XRparamvalue...)選擇參數(shù)設(shè)置tolToleranceondifferenceinfit1.0e4ma

2、xitersMaximumnumberofiterations50dimderdertoloopthroughdimensions1:ndims(A)initInitialguess[rom|nvecs|cellarray]printitnPrintfiteveryniterations0fnoprinting1[PU0out]=CP_ALS(...)alsoreturnsadditionaloutputthatcontainsthei

3、nputparameters.Note:The“fit“isdefinedas1nm(Xfull(P))nm(X)islooselytheproptionofthedatadescribedbytheCPmodeli.e.afitof1isperfect.%Examples:%X=sptenr([543]10)%P=cp_als(X2)%P=cp_als(X2dimder[321])%P=cp_als(X2dimder[321]init

4、nvecs)%U0=r(52)r(42)[]%InitialguessffactsofP%[PU0out]=cp_als(X2dimder[321]initU0)%P=cp_als(X2out.params)%Sameparamsaspreviousrun交替泊松回歸求張量X的非負(fù)CP分解cp_optFitsaCPmodeltoatensviaoptimization.cp_woptFitsaweightedCPmodeltoatens

5、viaoptimization.create_guessCreatesinitialguessfCPTuckerfitting.create_problemCreatetestproblemsftensfactizations.expt_dataExpttensrelateddatatoafile.impt_dataImpttensrelateddatatoafile.khatriraoKhatriRaoproductofmatrice

6、s.Y=khatrirao(AB)計(jì)算具有相同列數(shù)的矩陣A和B的khatrirao積[KRON(A(:1)B(:1))...KRON(A(:n)B(:n))]khatrirao(A1A2...)computestheKhatriRaoproductofmultiplematricesthathavethesamenumberofcolumns.khatrirao(C)computestheKhatriRaoproductofthemat

7、ricesincellarrayC.khatrirao(...r)computestheKhatriRaoproductinreverseder.%ExamplesA=r(52)B=r(32)C=r(22)khatrirao(AB)%KhatriRaoofABkhatrirao(BAr)%samethingasabovekhatrirao(CBA)%passingacellarraykhatrirao(ABCr)%sameasabove