1DGELS(1)              LAPACK driver routine (version 3.2)             DGELS(1)
2
3
4

NAME

6       DGELS  -  solves  overdetermined or underdetermined real linear systems
7       involving an M-by-N matrix A, or its transpose, using a QR or  LQ  fac‐
8       torization of A
9

SYNOPSIS

11       SUBROUTINE DGELS( TRANS,  M, N, NRHS, A, LDA, B, LDB, WORK, LWORK, INFO
12                         )
13
14           CHARACTER     TRANS
15
16           INTEGER       INFO, LDA, LDB, LWORK, M, N, NRHS
17
18           DOUBLE        PRECISION A( LDA, * ), B( LDB, * ), WORK( * )
19

PURPOSE

21       DGELS solves overdetermined  or  underdetermined  real  linear  systems
22       involving  an  M-by-N matrix A, or its transpose, using a QR or LQ fac‐
23       torization of A.  It is assumed that A has full  rank.   The  following
24       options are provided:
25       1. If TRANS = 'N' and m >= n:  find the least squares solution of
26          an overdetermined system, i.e., solve the least squares problem
27                       minimize || B - A*X ||.
28       2. If TRANS = 'N' and m < n:  find the minimum norm solution of
29          an underdetermined system A * X = B.
30       3. If TRANS = 'T' and m >= n:  find the minimum norm solution of
31          an undetermined system A**T * X = B.
32       4. If TRANS = 'T' and m < n:  find the least squares solution of
33          an overdetermined system, i.e., solve the least squares problem
34                       minimize || B - A**T * X ||.
35       Several right hand side vectors b and solution vectors x can be handled
36       in a single call; they are stored as the columns of the M-by-NRHS right
37       hand side matrix B and the N-by-NRHS solution matrix X.
38

ARGUMENTS

40       TRANS   (input) CHARACTER*1
41               = 'N': the linear system involves A;
42               = 'T': the linear system involves A**T.
43
44       M       (input) INTEGER
45               The number of rows of the matrix A.  M >= 0.
46
47       N       (input) INTEGER
48               The number of columns of the matrix A.  N >= 0.
49
50       NRHS    (input) INTEGER
51               The  number of right hand sides, i.e., the number of columns of
52               the matrices B and X. NRHS >=0.
53
54       A       (input/output) DOUBLE PRECISION array, dimension (LDA,N)
55               On entry, the M-by-N matrix A.  On exit, if M >= N, A is  over‐
56               written  by details of its QR factorization as returned by DGE‐
57               QRF; if M <  N, A is overwritten by details of its  LQ  factor‐
58               ization as returned by DGELQF.
59
60       LDA     (input) INTEGER
61               The leading dimension of the array A.  LDA >= max(1,M).
62
63       B       (input/output) DOUBLE PRECISION array, dimension (LDB,NRHS)
64               On  entry,  the  matrix  B  of  right hand side vectors, stored
65               columnwise; B is M-by-NRHS if TRANS  =  'N',  or  N-by-NRHS  if
66               TRANS  =  'T'.   On  exit, if INFO = 0, B is overwritten by the
67               solution vectors, stored columnwise: if TRANS = 'N' and m >= n,
68               rows  1  to  n of B contain the least squares solution vectors;
69               the residual sum of squares for the solution in each column  is
70               given  by  the sum of squares of elements N+1 to M in that col‐
71               umn; if TRANS = 'N' and m < n, rows 1 to N  of  B  contain  the
72               minimum  norm solution vectors; if TRANS = 'T' and m >= n, rows
73               1 to M of B contain the minimum norm solution vectors; if TRANS
74               =  'T'  and  m  < n, rows 1 to M of B contain the least squares
75               solution vectors; the residual sum of squares for the  solution
76               in  each  column is given by the sum of squares of elements M+1
77               to N in that column.
78
79       LDB     (input) INTEGER
80               The leading dimension of the array B. LDB >= MAX(1,M,N).
81
82       WORK      (workspace/output)   DOUBLE   PRECISION   array,    dimension
83       (MAX(1,LWORK))
84               On exit, if INFO = 0, WORK(1) returns the optimal LWORK.
85
86       LWORK   (input) INTEGER
87               The  dimension  of  the array WORK.  LWORK >= max( 1, MN + max(
88               MN, NRHS ) ).  For optimal performance, LWORK >= max( 1,  MN  +
89               max(  MN, NRHS )*NB ).  where MN = min(M,N) and NB is the opti‐
90               mum block size.  If LWORK =  -1,  then  a  workspace  query  is
91               assumed;  the  routine  only calculates the optimal size of the
92               WORK array, returns this value as the first entry of  the  WORK
93               array,  and  no  error  message  related  to LWORK is issued by
94               XERBLA.
95
96       INFO    (output) INTEGER
97               = 0:  successful exit
98               < 0:  if INFO = -i, the i-th argument had an illegal value
99               > 0:  if INFO =  i, the i-th diagonal element of the triangular
100               factor  of  A  is  zero, so that A does not have full rank; the
101               least squares solution could not be computed.
102
103
104
105 LAPACK driver routine (version 3.N2o)vember 2008                        DGELS(1)
Impressum