LAPACK  3.10.0
LAPACK: Linear Algebra PACKage

◆ dla_porpvgrw()

double precision function dla_porpvgrw ( character*1  UPLO,
integer  NCOLS,
double precision, dimension( lda, * )  A,
integer  LDA,
double precision, dimension( ldaf, * )  AF,
integer  LDAF,
double precision, dimension( * )  WORK 
)

DLA_PORPVGRW computes the reciprocal pivot growth factor norm(A)/norm(U) for a symmetric or Hermitian positive-definite matrix.

Download DLA_PORPVGRW + dependencies [TGZ] [ZIP] [TXT]

Purpose:
 DLA_PORPVGRW computes the reciprocal pivot growth factor
 norm(A)/norm(U). The "max absolute element" norm is used. If this is
 much less than 1, the stability of the LU factorization of the
 (equilibrated) matrix A could be poor. This also means that the
 solution X, estimated condition numbers, and error bounds could be
 unreliable.
Parameters
[in]UPLO
          UPLO is CHARACTER*1
       = 'U':  Upper triangle of A is stored;
       = 'L':  Lower triangle of A is stored.
[in]NCOLS
          NCOLS is INTEGER
     The number of columns of the matrix A. NCOLS >= 0.
[in]A
          A is DOUBLE PRECISION array, dimension (LDA,N)
     On entry, the N-by-N matrix A.
[in]LDA
          LDA is INTEGER
     The leading dimension of the array A.  LDA >= max(1,N).
[in]AF
          AF is DOUBLE PRECISION array, dimension (LDAF,N)
     The triangular factor U or L from the Cholesky factorization
     A = U**T*U or A = L*L**T, as computed by DPOTRF.
[in]LDAF
          LDAF is INTEGER
     The leading dimension of the array AF.  LDAF >= max(1,N).
[out]WORK
          WORK is DOUBLE PRECISION array, dimension (2*N)
Author
Univ. of Tennessee
Univ. of California Berkeley
Univ. of Colorado Denver
NAG Ltd.

Definition at line 104 of file dla_porpvgrw.f.

106 *
107 * -- LAPACK computational routine --
108 * -- LAPACK is a software package provided by Univ. of Tennessee, --
109 * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
110 *
111 * .. Scalar Arguments ..
112  CHARACTER*1 UPLO
113  INTEGER NCOLS, LDA, LDAF
114 * ..
115 * .. Array Arguments ..
116  DOUBLE PRECISION A( LDA, * ), AF( LDAF, * ), WORK( * )
117 * ..
118 *
119 * =====================================================================
120 *
121 * .. Local Scalars ..
122  INTEGER I, J
123  DOUBLE PRECISION AMAX, UMAX, RPVGRW
124  LOGICAL UPPER
125 * ..
126 * .. Intrinsic Functions ..
127  INTRINSIC abs, max, min
128 * ..
129 * .. External Functions ..
130  EXTERNAL lsame
131  LOGICAL LSAME
132 * ..
133 * .. Executable Statements ..
134 *
135  upper = lsame( 'Upper', uplo )
136 *
137 * DPOTRF will have factored only the NCOLSxNCOLS leading minor, so
138 * we restrict the growth search to that minor and use only the first
139 * 2*NCOLS workspace entries.
140 *
141  rpvgrw = 1.0d+0
142  DO i = 1, 2*ncols
143  work( i ) = 0.0d+0
144  END DO
145 *
146 * Find the max magnitude entry of each column.
147 *
148  IF ( upper ) THEN
149  DO j = 1, ncols
150  DO i = 1, j
151  work( ncols+j ) =
152  $ max( abs( a( i, j ) ), work( ncols+j ) )
153  END DO
154  END DO
155  ELSE
156  DO j = 1, ncols
157  DO i = j, ncols
158  work( ncols+j ) =
159  $ max( abs( a( i, j ) ), work( ncols+j ) )
160  END DO
161  END DO
162  END IF
163 *
164 * Now find the max magnitude entry of each column of the factor in
165 * AF. No pivoting, so no permutations.
166 *
167  IF ( lsame( 'Upper', uplo ) ) THEN
168  DO j = 1, ncols
169  DO i = 1, j
170  work( j ) = max( abs( af( i, j ) ), work( j ) )
171  END DO
172  END DO
173  ELSE
174  DO j = 1, ncols
175  DO i = j, ncols
176  work( j ) = max( abs( af( i, j ) ), work( j ) )
177  END DO
178  END DO
179  END IF
180 *
181 * Compute the *inverse* of the max element growth factor. Dividing
182 * by zero would imply the largest entry of the factor's column is
183 * zero. Than can happen when either the column of A is zero or
184 * massive pivots made the factor underflow to zero. Neither counts
185 * as growth in itself, so simply ignore terms with zero
186 * denominators.
187 *
188  IF ( lsame( 'Upper', uplo ) ) THEN
189  DO i = 1, ncols
190  umax = work( i )
191  amax = work( ncols+i )
192  IF ( umax /= 0.0d+0 ) THEN
193  rpvgrw = min( amax / umax, rpvgrw )
194  END IF
195  END DO
196  ELSE
197  DO i = 1, ncols
198  umax = work( i )
199  amax = work( ncols+i )
200  IF ( umax /= 0.0d+0 ) THEN
201  rpvgrw = min( amax / umax, rpvgrw )
202  END IF
203  END DO
204  END IF
205 
206  dla_porpvgrw = rpvgrw
207 *
208 * End of DLA_PORPVGRW
209 *
logical function lsame(CA, CB)
LSAME
Definition: lsame.f:53
double precision function dla_porpvgrw(UPLO, NCOLS, A, LDA, AF, LDAF, WORK)
DLA_PORPVGRW computes the reciprocal pivot growth factor norm(A)/norm(U) for a symmetric or Hermitian...
Definition: dla_porpvgrw.f:106
Here is the call graph for this function:
Here is the caller graph for this function: