2010 9 Sep., 2010 Journal On Numerical Methods and Computer Applications 31 3 Vol.31, No.3 Æ Æ Æ ! *1) ( , 510640; , ( 523808) , ( , ( 510640) , 100192) 510640) . , , . , , Moor-Penrose , : ; ; ; MR (2000) !"#: 65F18, 15A09 , , Kronecker n . " THE INVERSE GENERALIZED EIGENVALUE PROBLEM AND THE OPTIMAL APPROXIMATION FOR HERMITIAN-REFLEXIVE MATRICES Wang Jiangtao (Department of Mathematics, South China University of Technology, Guangzhou 510640, China; School of Business and administration, South China University of Technology, Guangzhou 510640, China) Zhang Zhongzhi (School of Computer Science and Technology, Dongguan University of Technology, Dongguan 523808, Guangdong, China) Xie Dongxiu (School of Sciences , Beijing Information Science and Technology University, Beijing 100192, China) Lei Xiuren (Department of Mathematics, South China University of Technology, Guangzhou 510640, China) * 1) 2010 ! "# 1 : 2 . (10971058), Æ$%&! (61N0810810). 3 233 Abstract The best approximation problem with the given spectral constraints is usually used to correct a stiffness and a mass matrix in the Vibration Control. Based on the denotative theorem of Hermitian-reflexive matrices, the author discuses the inverse generalized eigenvalue problem of Hermitian-reflexive and obtain the the general expression of the solution by using the straightened matrices and the Kronecker product of matrices. Furthermore, for any given complex matrices of dimension n, the expression of solution for its optimal approximate are presented by using the Moore-Penrose generalized inverse and the best approximation theory. Keywords: Hermitian-Reflexive Matrices; Generalized eigenvalue; Straightened matrix; Optimal approximation 2000 Mathematics Subject Classification: 65F18, 15A09 #$ 1. $%'&, '()Æ !"!#*(+!$)!#, ,$#*%*-+,&.!.'/0/("1.'/0)* (0 [1]). 1+, ,2#$!%!"&%&-.!- 2 )*. ω ,ω ,··· ,ω 1 2 m (m ≤ n) (m 3./3, φ1, φ2 , · · · , φm φ = (φ1 , φ2 , · · · , φm ), (04!%+. 4 2 Ω2 = diag(ω12 , · · · , ωm ). K &'-2!%&-., M &'-2!55-., 1676+,!7238: (1) ' /(): Kφ = M φΩ ; (2) *+8: K = K, M = M. Æ976%47238! K, M ,:9Æ9;:*+,-!.9 K, M !"1.'/0 )*. 5<6;/0, "'12!78, /(=<:9=,(3*-45%&-. K !6:7 '0 K >55-. M !6 : 0 M . 8? M.Baruch , ; (0 [2]) < $!@ = ÆA >? B C9((3, *--2 K > M , D Æ-. K, M @A76 2 T 0 T 0 0 0 min(K M ) − (K0 M0 ), A& K > M (7638 (1) > (2) !9. ,:E+FB;,-!%&12>5512, @,-;:(Æ!<: / 3>0 4!<: %+,Æ)**+,&Æ9-.!"1.'/0)*>> A0.!-.=G)*. ?9"1.'/0)*$>*?@*?A#CB?3H@,IA:B"C!4<, CD DDJEE*F!.G. .9.'/0)*>"1.'/0)*!HKIE4JÆ!LFM, 0 FG [3-9]. 1 +, FG [3] =<-.! QR 19$L7J! H N , HKJ I *+-."1 .'/0)*, 45)*9!ÆG:HJ; FG [5] =<KO!(3PÆ9"1'/0I)*; FG [7] K<LJ(3Æ9"1.'/0)*; FG [8] &=<-.!ML019>-.1Q 3HKJ*+2R*+-.!"1 '/0I)*. *-..'/0)*>"1.'/0 )* !HK, KE(L<-.! QR 19?ML019,(3, MFS=<-.TM,(3PNU NOV'3I-.!"1.'/0)*. WONWOXO 2010 PQ, DRÆSMF&@<!PY>QT. 4 C :U m × n +R-.Z[, R : U m×n +I-.Z[,I :U k \SV-., SR , ASR 1B:U n \*+-.>I* +-.!Z[. HC , AHC 1B:U n \NOV'-.>INOV'-.Z[,U C :U n \P-.Z[. A , A 1B:U A ! Moore-Penrose "1.>TQRS, tr(A) :U A !], *9-. A, B ∈ C , A⊗B ∈ C :U A > B ! Kronecker ^, < A, B > : U A > B !_^, A;1& < A, B >= tr(B A), ?U_^TV!W #& 234 m×n m×n n×n k n×n n×n n×n + n×n H m×n mp×nq H A = < A, A > = tr(AH A). W., %4W#( Frobenius W#, CU C -FÆ XX!_^`a. · :UY5 ! 2 W#.GRC :U n \"1IZ-.UF!Z[, D n×n 2 n×n V % A = (a 1. GRC n×n = {P |P H = P, P 2 = In , P ∈ C n×n }. ij )m×n , b ai = (a1i , a2i, · · · , ami)T (i = 1, 2, · · · , n). 4 vec(A) = (aT1 , aT2 , · · · , aTn )T . +Y5 vec(A) &-. A !TM (YL). +M: AT = A, b a1 = (a11, a21, · · · , an1)T , a2 = (a22, a32, · · · , an2)T , · · · , an−1 = (a(n−1)(n−1) , an(n−1) )T , an = ann . Kb vecS (A) = (aT1 , aT2 , · · · , aTn−1 , aTn )T ∈ R & n(n+1) 2 K?-.TM!;1>FG [10] &!(3, WX4,2D). Y 1. * X ∈ Rn×n, K,2cJFd . X ∈ SRn×n ⇐⇒ vec(X) = Kn vecS (X), (1) A& vec (X) ∈ R S ⎛ ⎜ ⎜ ⎜ ⎜ ⎜ Kn = ⎜ ⎜ ⎜ ⎜ ⎝ n(n+1) 2 , Kn ?,J#$,ei &SV-. In !Z i L, e1 e2 e3 ··· en−1 en 0 0 ··· 0 0 ··· 0 0 0 e1 0 ··· 0 0 e2 e3 ··· en−1 en ··· 0 0 0 .. . 0 0 .. . 0 e1 .. . 0 ··· 0 .. . 0 0 .. . 0 e2 .. . 0 ··· .. . ··· 0 .. . e2 0 .. . 0 ··· ··· ··· 0 .. . e1 ··· ··· 0 .. . en−1 0 .. . en 0 0 0 ··· 0 e1 0 0 ··· 0 e2 ··· 0 en−1 0 ⎞ ⎟ 0 ⎟ ⎟ 0 ⎟ ⎟ . .. ⎟ . ⎟ ⎟ ⎟ 0 ⎠ en (1.1) (2) X ∈ ASRn×n ⇐⇒ vec(X) = Tn vecS (X), vecS (X), ei , Tn , ⎛ 0 e2 e3 · · · en−1 en 0 0 ··· ⎜ 0 ··· 0 0 0 e3 · · · ⎜ 0 −e1 ⎜ ⎜ 0 0 0 0 −e2 · · · 0 −e1 · · · ⎜ Tn = ⎜ . . . . . .. .. .. .. .. .. .. ⎜ .. ··· . . . ⎜ ⎜ ⎝ 0 0 0 · · · −e1 0 0 0 ··· A& K% 0 0 0 ?,J#$ ··· 0 −e1 0 0 ··· 0 0 ··· 0 0 en−1 0 .. . en 0 .. . ··· ··· ··· 0 0 .. . 0 0 .. . −e2 0 ··· 0 en 0 −e2 ··· 0 −en−1 0 ⎞ ⎟ ⎟ ⎟ ⎟ ⎟ ⎟. ⎟ ⎟ ⎟ 0 ⎠ 0 0 .. . 0 (1.2) 3 V % 235 2. #; P ∈ GRC n×n, + A ∈ C n×n &.9 P !NOV'3I-., +M176: AH = A, P AP = A. b[: n \NOV'3I-.!Z[& HCrn×n (P ) ( r :U3I-.), D HCrn×n (P ) = {A|P AP = A, AH = A, A ∈ C n×n }. W., -.Z[ HC (P ) > P !ZE:., MF[e;-. P (<;[\!. 9(, NOV'3I-."1.'/0)*f0.=G)* !#*g4+,: n×n r \]. #; X ∈ C n×m , Λ = diag(λ1, λ2 , · · · , λm ) ∈ C m×m, 4 SA,B = (A, B)|AX = BXΛ, A, B ∈ HCrn×n (P ) . (1.3) *#;!-. A∗ , B∗ ∈ C n×n , Æ A, B ∈ SA,B @4 (A∗ B) = B ∗ ) − (A min (A,B)∈SA,B (A∗ B ∗ ) − (A B). (1.4) MF,-+,: $Z 2 h, XE)*9!]$8>\Æ8, ^]VA\Æ94:HJ; Z 3 h, #$Æ9)*!#0^3-f#0^1. 2. '() ^_`ab c d c 4 P1 = 1 1 (I + P ), P2 = (I − P ), 2 2 [iXE P1 , P2 j&2R_e-., D`: P1 + P2 = I, P1P2 = 0. CU, ]$SVL2 R-. Q1 ∈ C n×r , Q2 ∈ C n×(n−r), @4 P1 = Q1QH1 , P2 = Q2QH2 , a4 Q = (Q1, Q2) ∈ C n×n, K Q ∈ U C n×n. Y 2[11] . 4 & S= & M1 A|A = Q 0 0 M2 H Q , M1 ∈ HC r×r , M2 ∈ HC 9(: S = HCrn×n(P ). Y 3[12] . A ∈ C m×n, b ∈ C m , (1) b 8()U Ax = b :9!_( 3 8 ( Uc, 59& (2) AA+ b = b. x = A+ b + (I − A+ A)y, ∀y ∈ C n . x = A+ b + (I − A+ A)y, ∀y ∈ C n . [0db8()U Ax = b !f=`a9& (n−r)×(n−r) . WONWOXO 236 2010 *eg! M1, N1 ∈ HC r×r , M2, N2 ∈ HC (n−r)×(n−r), :,JFd M1 = M11 + iM12 , N1 = N11 + iN12 , (2.1) M2 = M21 + iM22 , N2 = N21 + iN22 . (2.2) A& M ∈ SRr×r , M12 , N12 ∈ ASRr×r , M21 , N21 ∈ SR(n−r)×(n−r), M22 , N22 ∈ ASR . 11 , N11 (n−r)×(n−r) ?D) 1 ?D) 2 -fFG [8] !fg, [iXE,2;). VY 1. #; X ∈ C n×m, Λ = diag(λ1, λ2, · · · , λm) ∈ C m×m, `4 H Q X= X1 X2 , (2.3) R1 = (X1T ⊗ Ir )Kr , R2 = (X1T ⊗ Ir )iTr , R3 = ((X1 Λ)T ⊗ Ir )Kr , R4 = ((X1 Λ)T ⊗ Ir )iTr , W1 = diag(Kr Tr Kr Tr ), (2.4) G1 = (R1 R2 −R3 −R4 ), R5 = (X2T ⊗ In−r )Kn−r , R6 = (X2 ⊗ In−r )iTn−r , R7 = ((X2 Λ)T ⊗ In−r )Kn−r , R8 = ((X2 Λ)T ⊗ In−r )iTn−r , G2 = (R5 R6 −R7 −R8 ), W2 = diag(Kn−r Tn−r Kn−r Tn−r ). (2.5) KZ[ SA,B b`, `A=h*:U&: M1 A=Q 0 0 QH , M2 N1 B=Q 0 0 N2 QH , (2.6) ,C Mi, Ni , i = 1, 2 + (2.1), (2.2) J;1, `: ⎛ ⎞ vec(M11 ) ⎜ ⎟ ⎜vec(M12 )⎟ + ⎜ ⎟ ⎜ vec(N ) ⎟ = W1 (I − G1 G1 )x, 11 ⎠ ⎝ vec(N12 ) (2.7) ⎛ ⎞ vec(M21 ) ⎜ ⎟ ⎜vec(M22 )⎟ + ⎜ ⎟ ⎜ vec(N ) ⎟ = W2 (I − G2 G2 )y, 21 ⎠ ⎝ vec(N22 ) (2.8) A& x ∈ C , y ∈ C &egY5. ,2, XE)*9!]$8>iÆ8c^]Vfk=G9!:HJ. VY 2. *#;-. A , B ∈ C , A1PY>;) 1 0K c`4 2r(r+1) 2(n−r)(n−r+1) ∗ ∗ ∗ ∗ QH 1 A Q1 = A1 + iA2 , ∗ ∗ ∗ QH 2 B Q2 = B3 + iB4 , ∗ n×n ∗ ∗ ∗ QH 2 A Q2 = A3 + iA4 , ∗ ∗ ∗ QH 1 B Q1 = B1 + iB2 , Ai , Bi ∈ Rn×n , i = 1, 2, 3, 4. K)*]$\Æ!fk=G9, `, 9*:U&: =Q A M1 0 0 M2 H Q , =Q B (2.9) N1 0 0 N2 QH , (2.10) A& M , N , i =⎛1, 2 + (2.1),⎞ (2.2) J;1, `: 237 3 i i ⎞ ⎛ vec(M11 ) vec(A∗1 ) ⎟ ⎟ ⎜ ⎜ ∗ ⎟ ⎜vec(M12 )⎟ ⎜ ⎟ = W1 (I − G+ G1 )(W1 (I − G+ G1 ))+ ⎜vec(A2 ) ⎟ , ⎜ 1 1 ⎜ vec(N ) ⎟ ⎜vec(B ∗ )⎟ 11 ⎠ ⎝ ⎝ 1 ⎠ vec(N12 ) vec(B2∗ ) ⎛ ⎞ ⎞ ⎛ vec(M21 ) vec(A∗3 ) ⎜ ⎟ ⎟ ⎜ ∗ ⎟ ⎜vec(M22 )⎟ ⎜ ⎜ ⎟ = W2 (I − G+ G2 )(W2 (I − G+ G2 ))+ ⎜vec(A4 ) ⎟ . 2 2 ⎜ vec(N ) ⎟ ⎜vec(B ∗ )⎟ 21 ⎠ ⎝ ⎝ 3 ⎠ vec(N22 ) vec(B4∗ ) *? J*h )*&!Z[ K?fk=G;)*h )*]$\Æ!fk=G9 . (2.6), (2.7), (2.8) , , 2P]V, , SA,B . (`a C n×n × C n×n (2.11) (2.12) &!djZ, fk=G9. *#;! A∗, B∗ ∈ C n×n, A, B ∈ SA,B , `? (2.9) J*h 2 2 2 f 2 (A∗ B ∗ ) − (A B) = A∗ − A + B ∗ − B 2 2 0 M1 N ∗ 1 = A − Q QH + B ∗ − Q QH 0 N2 0 M2 2 2 0 M1 N1 0 = QH A∗ Q − + QH B ∗ Q − 0 M2 0 N2 2 QH A∗ Q QH A∗ Q QH B ∗ Q QH B ∗ Q M1 N1 0 1 2 1 2 1 1 1 1 = − − + ∗ H ∗ H ∗ H ∗ QH 0 M 0 A Q Q A Q Q B Q Q B Q 1 2 2 1 2 2 2 2 2 H ∗ 2 H ∗ 2 H ∗ 2 H ∗ 2 = Q1 A Q1 − M1 + Q2 A Q2 − M2 + Q1 A Q2 + Q2 A Q1 2 H ∗ ∗ + Q B Q2 − N2 2 + QH B ∗ Q2 2 + QH B ∗ Q1 2 + QH 1 B Q1 − N1 2 1 2 2 2 2 ∗ = (A∗1 − M11 ) + i(A∗2 − M12 ) + (A∗3 − M21 ) + i(A∗4 − M22 ) + QH 1 A Q2 H ∗ 2 ∗ 2 2 + Q2 A Q1 + (B1 − N11 ) + i(B2∗ − N12 ) + (B3∗ − N21 ) + i(B4∗ − N22 ) 2 H ∗ 2 ∗ + Q2 B Q1 . + QH 1 B Q2 ? F W#!85-f F W#> 2 W#!.!*45 2 N2 2 2 2 2 2 ∗ f 2 = A∗1 − M11 + A∗2 − M12 + A∗3 − M21 + A∗4 − M22 + QH 1 A Q2 H ∗ 2 ∗ 2 ∗ 2 ∗ 2 ∗ 2 + Q2 A Q1 + B1 − N11 + B2 − N12 + B3 − N21 + B4 − N22 2 H ∗ 2 ∗ + Q B Q1 + QH 1 B Q2 2 2 2 2 2 ∗ = vec(A1 − M11 ) 2 + vec(A∗2 − M12 )2 + vec(A∗3 − M21 )2 + vec(A∗4 − M22 )2 2 2 2 2 + vec(B1∗ − N11 )2 + vec(B2∗ − N12 )2 + vec(B3∗ − N21 )2 + vec(B4∗ − N22 )2 2 H ∗ 2 H ∗ 2 H ∗ 2 ∗ + Q2 A Q1 + Q1 B Q2 + Q2 B Q1 + QH 1 A Q2 ⎛ ⎛ ⎞ ⎞ vec(A∗ − M ) 2 vec(A∗ − M ) 2 11 21 1 3 ⎟ ⎟ ⎜ ⎜ H ∗ 2 H ∗ 2 ⎜vec(A∗2 − M12 )⎟ ⎜vec(A∗4 − M22 )⎟ ⎟ ⎟ ⎜ ⎜ = ⎜ vec(B ∗ − N ) ⎟ + ⎜ vec(B ∗ − N ) ⎟ + Q1 A Q2 + Q2 A Q1 11 ⎠ 21 ⎠ ⎝ ⎝ 1 3 vec(B2∗ − N12 ) vec(B4∗ − N22 ) 2 2 2 H ∗ 2 ∗ + QH B Q + B Q Q 2 1 1 2 0 238 WONWOXO 2010 ⎛ ⎞ ⎛ ⎞2 ⎛ ⎞ ⎛ ⎞2 vec(A∗ ) vec(A∗ ) vec(M vec(M ) ) 11 21 1 3 ⎜ ⎜ ⎜ ⎜ ⎟ ⎟ ⎟ ⎟ ⎜vec(A∗2 ) ⎟ ⎜vec(M12 )⎟ ⎜vec(A∗4 ) ⎟ ⎜vec(M22 )⎟ ⎜ ⎜ ⎜ ⎜ ⎟ ⎟ ⎟ ⎟ −⎜ = ⎜ ⎟ + ⎜vec(B ∗ )⎟ − ⎜ vec(N ) ⎟ ∗ ⎟ ⎝vec(B1 )⎠ ⎝ vec(N11 ) ⎠ ⎝ 21 ⎠ ⎝ 3 ⎠ vec(B2∗ ) vec(B4∗ ) vec(N12 ) vec(N22 ) 2 2 H ∗ 2 2 H ∗ 2 H ∗ 2 ∗ + Q2 A Q1 + Q1 B Q2 + Q2 B Q1 + QH 1 A Q2 ⎛ 2 ⎛ 2 ⎞ ⎞ vec(A∗ ) vec(A∗ ) 1 3 ⎜ ⎜ ⎟ ⎟ ∗ ∗ ⎜ ⎜vec(A2 ) ⎟ ⎟ vec(A ) 4 ⎟ + + ⎜ ⎜ ⎟ − W1 (I − G1 G1 )x + ⎜ − W2 (I − G2 G2 )y = ⎜ ∗ ⎟ ∗ ⎟ ⎝vec(B1 )⎠ ⎝vec(B3 )⎠ vec(B2∗ ) vec(B4∗ ) 2 2 H ∗ 2 H ∗ 2 H ∗ 2 H ∗ 2 + Q1 A Q2 + Q2 A Q1 + Q1 B Q2 + Q2 B Q1 9(, ? (2.13) J*h, i)* (A,B)∈S min ∗ A,B (A ∗ B ) − (A B) ,:9,2c ⎛ ⎞ vec(A∗ ) 1 ⎜ ⎟ ⎜vec(A∗2 ) ⎟ + ⎟ ⎜ min ⎜ − W1 (I − G1 G1 )x , x vec(B ∗ )⎟ ⎝ 1 ⎠ vec(B2∗ ) 2 ⎛ ⎞ vec(A∗ ) 3 ⎜ ⎟ ⎜vec(A∗4 ) ⎟ + ⎜ ⎟ − W2 (I − G2 G2 )y min ⎜ . y vec(B ∗ )⎟ ⎝ 3 ⎠ vec(B4∗ ) K?D) 3(2) *h, (2.14), (2.15) J!9*1B:U& )* (2.13) (2.14) (2.15) 2 + + x = x0 + (I − (W1 (I − G+ 1 G1 )) (W1 (I − G1 G1 )))x1 , ⎛ ⎞ vec(A∗1 ) ⎜ ⎟ ∗ ⎟ ⎜ + + ⎜ vec(A2 )⎟ . x0 = (W1 (I − G1 G1 )) ⎜ ∗ ⎟ ⎝vec(B1 )⎠ vec(B2∗ ) + + y = y0 + (I − (W2 (I − G+ 2 G2 )) (W2 (I − G2 G2 )))y1 , ⎛ ⎞ vec(A∗3 ) ⎜ ⎟ ∗ ⎟ ⎜ + + ⎜ vec(A4 ) ⎟ . y0 = (W2 (I − G2 G2 )) ⎜ ∗ ⎟ ⎝vec(B3 )⎠ vec(B4∗ ) (2.16) (2.17) ,C x1 ∈ C 2r(r+1), y1 ∈ C 2(n−r)(n−r+1) (e g! Y 5. 1BS (2.16), (2.17) J e R (2.7), (2.8) J D*45 (1.4) J !9. Xf. (+, jk l m ?;) 2 *h, *4Æ9fk=G)*9!^3+,: #; P, A∗ .B∗, X, Λ. (1) Y (2.9) JÆ $ A∗1 , A∗2 , A∗3 , A∗4 , B1∗ , B2∗ , B3∗ , B4∗ , 3. 3 239 Y (1.1),(1.2) JÆ$ Kr , Kn−r , Tr , Tn−r , (3) ? X, Y (2.3) JÆ $ X1 , X2 , (4) Y (2.4) J ' ^ $ R1 , R2 , R3 , R4 , G1 , W1 , Y (2.5) J ' ^ $ R5 , R6 , R7 , R8 , G2 , W2 , (5) Y (2.11) JÆ 4 (vec(M11 ) vec(M12 ) vec(N11 ) vec(N12 ))T , Y (2.12) JÆ 4 (vec(M21 ) vec(M22 ) vec(N21 ) vec(N22 ))T c (6) 1BS Æ 4! Y 5li, . mY (2.1), (2.2) J 45-. M1 , M2 , N1 , N2 , fmne R B. (2.10) JÆ 4 )* & (1.4) J !fk=G0 A, o. ,CE n = 4, r = 2, m = 4, `#; (2) ⎞ −0.8944 − 0.4472i 0 ⎟ ⎜ ⎜ 0 −0.8944 − 0.4472i⎟ ⎟, ⎜ P =⎜ ⎟ 0 0 0 ⎠ ⎝−0.8944 + 0.4472i 0 −0.8944 + 0.4472i 0 0 ⎛ 0 0 0 0 ⎞ 1.0000 0 + 2.0000i 1.0000 2.0000 ⎟ ⎜ ⎜ 0 1.0000 1.0000 0 + 2.0000i⎟ ∗ ⎟, ⎜ A =⎜ 0 + 1.0000i 2.0000 0 + 1.0000i⎟ ⎠ ⎝ 0 2.0000 1.0000 0 + 1.0000i 1.0000 ⎞ ⎛ 1.0000 0 + 1.0000i 2.0000 2.0000 ⎟ ⎜ ⎜ 1.0000 0 0 + 3.0000i 2.0000⎟ ∗ ⎟, ⎜ B =⎜ 2.0000 1.0000 0 ⎟ ⎠ ⎝0 + 1.0000i 1.0000 1.0000 0 + 3.0000i 2.0000 ⎛ ⎞ 1.0000 2.0000 + 2.0000i 0 + 2.0000i 3.0000 + 1.0000i ⎜ ⎟ ⎜3.0000 + 2.0000i 2.0000 + 2.0000i 2.0000 + 1.0000i 1.0000 + 2.0000i⎟ ⎜ ⎟, X=⎜ ⎟ 0 + 1.0000i 0 1.0000 ⎝ 0 + 2.0000i ⎠ 3.0000 + 2.0000i 1.0000 2.0000 2.0000 ⎛ ⎞ 1 0 0 0 ⎜ ⎟ ⎜0 1 0 0⎟ ⎜ ⎟. (1.1), (1.2) Kr , Kn−r , Tr , Tn−r Λ=⎜ ⎟ ⎝0 0 2 0⎠ 0 0 0 2 ⎛ JÆ$! Y ⎛ 1 0 ⎜ ⎜0 1 K2 = ⎜ ⎜0 1 ⎝ 0 0 Y%4gp45lim!9 & 0.1056 1.1736 M11 = N12 = 1.1736 1.0528 0 0.2854 , −0.2854 0 M12 = , M21 = ⎞ 0 ⎟ 0⎟ ⎟, 0⎟ ⎠ 1 1B& ⎛ 0 0 0 ⎞ ⎜ ⎟ ⎜0 1 0 ⎟ ⎜ ⎟. T2 = ⎜ ⎟ ⎝0 −1 0⎠ 0 0 0 0 −0.0972 1.9472 −0.7264 0.0972 0 , −0.7264 1.8944 N11 = , M22 −0.3416 1.2028 , 1.2028 0.3239 0 −0.7972 = , 0.7972 0 WONWOXO 240 1.6708 −2.0972 , −2.0972 2.3416 N21 = N22 = hD45 2010 0 0.3854 . −0.3854 0 0.1056 − 0.0000i 1.1736 + 0.0972i 1.9472 + 0.0000i −0.7264 − 0.7972 M1 = , M2 = , 1.1736 − 0.0972i 1.0528 − 0.0000i −0.7264 + 0.7972i 1.8944 − 0.0000i −0.3416 − 0.0000i 1.2028 − 0.2854i 1.6708 + 0.0000i −2.0972 + 0.3854i N1 = , N2 = , 1.2028 + 0.2854i 0.3292 − 0.0000i −2.0972 − 0.3854i 2.3416 + 0.0000i fq45 ⎛ )* !fk=G9 & i 0.5000 + 0.0000 ⎜ ⎜0.0000 − 0.5000i =⎜ A ⎜0.4000 − 0.2000i ⎝ 0.9500 − 0.3500i ⎛ 1.0000 + 0.0000i ⎜ ⎜ = ⎜0.2500 + 0.5000i B ⎜0.6000 − 0.3000i ⎝ 1.6500 + 0.0500i [1] ⎞ 0.9500 + 0.3500i ⎟ 0.8000 + 0.4000i ⎟ ⎟, 1.5000 + 0.0000i −0.0000 + 0.5000i⎟ ⎠ 0.0000 − 0.5000i 1.0000 − 0.0000i ⎞ 0.6000 + 0.3000i 1.6500 − 0.0500i ⎟ 0.9500 + 1.3500i 1.2000 + 0.6000i⎟ ⎟. 1.000 + 0.0000i 0.2500 − 0.5000i⎟ ⎠ 0.2500 + 0.5000i 1.0000 − 0.0000i −0.0000 + 0.5000i 0.4000 + +0.2000i 1.0000 − 0.0000i 0.8500 + 0.5500i 0.8500 − 0.5500i 0.8000 − 0.4000i 0.2500 − 0.5000i 1.0000 − 0.0000i 0.9500 − 1.3500i 1.2000 − 0.6000i r s t u jkl, kvn, olm. mp, Oqnrowxs. n" p o t, pl, 1986 8 . [2] Baruch M. Optimization procedure to correct stiffness and flexibility matrices using vibration q. q rr [J]. WO, 1990, 2: 177-187. uss, vty, ow. qrxr [J]. WONWO XO, tests[J]. AIAA J, 1978, 16: 1208-1210. [3] [4] uzv. wytu [J]. WO, 1993, 3: 310-317. 2007, 28(4): 272-289. [5] [6] Dai H. An algorithm for symmetric generalized inverse eigenvalue problems[J]. Linear Algebra Appl., 1999, 296: 79-98. [7] Dai H, Lancaster P. Newton’s method for a generalized inverse eigenvalue problems[J]. Number, {x, yvv. rxr [J]. z{w (x), 2006, 44(3): 185- Linear Algebra Appl., 1997, 4: 1-21. [8] 188. [9] Zhang Z Z, Hu X Y and Zhang L. The solvability conditions for the inverse eigenvalue problem |zy, |l{, }}. tz AXB + CY D = E r~|{||} [J]. WO, 2007, of Herimitian-generallized Hamiltonian matrices[J]. Inverse Problems, 2002, 18: 1369-1376. [10] 29(2): 203-215. [11] Lijun Zhao, Xiyan Hu, Lei Zhang. Linear restriction problem of Hermitian reflexive matrices and {}~, q. ~ [M]. : ~x, 1991. its approximation[J]. Applied Mathematics and Computation, 2008, 200: 341-351. [12]