概要
本サンプルはFortran言語によりLAPACKルーチンZGGEVXを利用するサンプルプログラムです。
行列対

及び

条件数の推定値とそれぞれの固有値と固有ベクトルの前方誤差限界も合わせて求めます。行列対を均衡化するオプションが使用されます。
入力データ
(本ルーチンの詳細はZGGEVX のマニュアルページを参照)このデータをダウンロード |
ZGGEVX Example Program Data 4 : Value of N (-21.10,-22.50) ( 53.50,-50.50) (-34.50,127.50) ( 7.50, 0.50) ( -0.46, -7.78) ( -3.50,-37.50) (-15.50, 58.50) (-10.50, -1.50) ( 4.30, -5.50) ( 39.70,-17.10) (-68.50, 12.50) ( -7.50, -3.50) ( 5.50, 4.40) ( 14.40, 43.30) (-32.50,-46.00) (-19.00,-32.50) : End of A ( 1.00, -5.00) ( 1.60, 1.20) ( -3.00, 0.00) ( 0.00, -1.00) ( 0.80, -0.60) ( 3.00, -5.00) ( -4.00, 3.00) ( -2.40, -3.20) ( 1.00, 0.00) ( 2.40, 1.80) ( -4.00, -5.00) ( 0.00, -3.00) ( 0.00, 1.00) ( -1.80, 2.40) ( 0.00, -4.00) ( 4.00, -5.00) : End of B
出力結果
(本ルーチンの詳細はZGGEVX のマニュアルページを参照)この出力例をダウンロード |
Warning: Floating underflow occurred ZGGEVX Example Program Results Eigenvalues Eigenvalue 1 ( 3.0000E+00,-9.0000E+00) 2 ( 4.0000E+00,-5.0000E+00) 3 ( 2.0000E+00,-5.0000E+00) 4 ( 3.0000E+00,-1.0000E+00) Eigenvectors Eigenvector 1 ( 1.0000E+00, 0.0000E+00) ( 1.6000E-01,-1.2000E-01) ( 1.2000E-01, 1.6000E-01) (-1.6000E-01, 1.2000E-01) 2 ( 1.0000E+00, 0.0000E+00) ( 8.8889E-03,-6.6667E-03) (-3.3333E-02, 1.1796E-16) ( 4.1633E-16, 1.5556E-01) 3 ( 1.0000E+00, 0.0000E+00) ( 4.5714E-03,-3.4286E-03) ( 6.2857E-02,-1.4572E-16) ( 1.4572E-16, 6.2857E-02) 4 ( 1.0000E+00, 0.0000E+00) ( 1.6000E-01,-1.2000E-01) ( 1.2000E-01,-1.6000E-01) ( 1.6000E-01, 1.2000E-01)
ソースコード
(本ルーチンの詳細はZGGEVX のマニュアルページを参照)※本サンプルソースコードのご利用手順は「サンプルのコンパイル及び実行方法」をご参照下さい。
このソースコードをダウンロード |
Program zggevx_example ! ZGGEVX Example Program Text ! Copyright 2017, Numerical Algorithms Group Ltd. http://www.nag.com ! .. Use Statements .. Use lapack_example_aux, Only: nagf_sort_realvec_rank_rearrange, & nagf_blas_dpyth, nagf_sort_cmplxvec_rank_rearrange, & nagf_sort_realvec_rank Use lapack_interfaces, Only: zggevx Use lapack_precision, Only: dp ! .. Implicit None Statement .. Implicit None ! .. Parameters .. Integer, Parameter :: nb = 64, nin = 5, nout = 6 Logical, Parameter :: verbose = .False. ! .. Local Scalars .. Complex (Kind=dp) :: eig, scal Real (Kind=dp) :: abnorm, abnrm, bbnrm, eps, small, tol Integer :: i, ifail, ihi, ilo, info, j, k, lda, ldb, ldvr, lwork, n ! .. Local Arrays .. Complex (Kind=dp), Allocatable :: a(:, :), alpha(:), b(:, :), beta(:), & temp(:), vr(:, :), work(:) Complex (Kind=dp) :: dummy(1, 1) Real (Kind=dp), Allocatable :: lscale(:), rconde(:), rcondv(:), & rscale(:), rwork(:) Integer, Allocatable :: irank(:), iwork(:) Logical, Allocatable :: bwork(:) ! .. Intrinsic Procedures .. Intrinsic :: abs, epsilon, max, maxloc, nint, real, tiny ! .. Executable Statements .. Write (nout, *) 'ZGGEVX Example Program Results' ! Skip heading in data file Read (nin, *) Read (nin, *) n lda = n ldb = n ldvr = n Allocate (a(lda,n), alpha(n), b(ldb,n), beta(n), vr(ldvr,n), lscale(n), & rconde(n), rcondv(n), rscale(n), rwork(6*n), iwork(n+2), bwork(n), & temp(n)) ! Use routine workspace query to get optimal workspace. lwork = -1 Call zggevx('Balance', 'No vectors (left)', 'Vectors (right)', & 'Both reciprocal condition numbers', n, a, lda, b, ldb, alpha, beta, & dummy, 1, vr, ldvr, ilo, ihi, lscale, rscale, abnrm, bbnrm, rconde, & rcondv, dummy, lwork, rwork, iwork, bwork, info) ! Make sure that there is enough workspace for block size nb. lwork = max((nb+2*n)*n, nint(real(dummy(1,1)))) Allocate (work(lwork)) ! Read in the matrices A and B Read (nin, *)(a(i,1:n), i=1, n) Read (nin, *)(b(i,1:n), i=1, n) ! Solve the generalized eigenvalue problem Call zggevx('Balance', 'No vectors (left)', 'Vectors (right)', & 'Both reciprocal condition numbers', n, a, lda, b, ldb, alpha, beta, & dummy, 1, vr, ldvr, ilo, ihi, lscale, rscale, abnrm, bbnrm, rconde, & rcondv, work, lwork, rwork, iwork, bwork, info) If (info>0) Then Write (nout, *) Write (nout, 100) 'Failure in ZGGEVX. INFO =', info Else ! Compute the machine precision, the safe range parameter ! SMALL and sqrt(ABNRM**2+BBNRM**2) eps = epsilon(1.0E0_dp) small = tiny(1.0E0_dp) abnorm = nagf_blas_dpyth(abnrm, bbnrm) tol = eps*abnorm ! Reorder eigenvalues by descending absolute value rwork(1:n) = abs(alpha(1:n)/beta(1:n)) Allocate (irank(n)) ifail = 0 Call nagf_sort_realvec_rank(rwork, 1, n, 'Descending', irank, ifail) Call nagf_sort_cmplxvec_rank_rearrange(alpha, 1, n, irank, ifail) Call nagf_sort_cmplxvec_rank_rearrange(beta, 1, n, irank, ifail) Call nagf_sort_realvec_rank_rearrange(rconde, 1, n, irank, ifail) ! Reorder eigenvectors accordingly Do j = 1, n temp(1:n) = vr(j, 1:n) Call nagf_sort_cmplxvec_rank_rearrange(temp, 1, n, irank, ifail) vr(j, 1:n) = temp(1:n) End Do Call nagf_sort_realvec_rank_rearrange(rcondv, 1, n, irank, ifail) ! Print out eigenvalues and vectors and associated condition ! number and bounds Write (nout, *) Write (nout, *) 'Eigenvalues' Write (nout, *) If (verbose) Then Write (nout, *) ' Eigenvalue rcond error' Else Write (nout, *) ' Eigenvalue' End If Do j = 1, n ! Print out information on the j-th eigenvalue If ((abs(alpha(j)))*small>=abs(beta(j))) Then If (rconde(j)>0.0_dp) Then If (tol/rconde(j)<500.0_dp*eps) Then Write (nout, 140) j, rconde(j), '-' Else Write (nout, 150) j, rconde(j), tol/rconde(j) End If Else Write (nout, 140) j, rconde(j), 'Inf' End If Else eig = alpha(j)/beta(j) If (verbose) Then If (rconde(j)>0.0_dp) Then If (tol/rconde(j)<500.0_dp*eps) Then Write (nout, 110) j, eig, rconde(j), '-' Else Write (nout, 120) j, eig, rconde(j), tol/rconde(j) End If Else Write (nout, 110) j, eig, rconde(j), 'Inf' End If Else Write (nout, 110) j, eig End If End If End Do Write (nout, *) Write (nout, *) 'Eigenvectors' Write (nout, *) If (verbose) Then Write (nout, *) ' Eigenvector rcond error' Else Write (nout, *) ' Eigenvector' End If Do j = 1, n ! Print information on j-th eigenvector Write (nout, *) ! Re-normalize eigenvector, largest absolute element real (=1) rwork(1:n) = abs(vr(1:n,j)) k = maxloc(rwork(1:n), 1) scal = (1.0_dp, 0.0_dp)/vr(k, j) vr(1:n, j) = vr(1:n, j)*scal If (verbose) Then If (rcondv(j)>0.0_dp) Then If (tol/rcondv(j)<500.0_dp*eps) Then Write (nout, 110) j, vr(1, j), rcondv(j), '-' Else Write (nout, 120) j, vr(1, j), rcondv(j), tol/rcondv(j) End If Else Write (nout, 110) j, vr(1, j), rcondv(j), 'Inf' End If Else Write (nout, 110) j, vr(1, j) End If Write (nout, 130) vr(2:n, j) End Do If (verbose) Then Write (nout, *) Write (nout, *) & 'Errors below 500*machine precision are not displayed' End If End If 100 Format (1X, A, I4) 110 Format (1X, I2, 1X, '(', 1P, E11.4, ',', E11.4, ')', 1X, 0P, F7.4, 4X, & A) 120 Format (1X, I2, 1X, '(', 1P, E11.4, ',', E11.4, ')', 1X, 0P, F7.4, 1X, & 1P, E8.1) 130 Format (1X, 3X, '(', 1P, E11.4, ',', E11.4, ')') 140 Format (1X, I2, 1X, ' Infinite or undetermined', 1X, 0P, F7.4, 4X, A) 150 Format (1X, I2, 1X, ' Infinite or undetermined', 1X, 0P, F7.4, 1X, 1P, & E8.1) End Program