1
2      SUBROUTINE LPDP(A,MDA,M,N1,N2,PRGOPT,X,WNORM,MODE,WS,IS)
3C***BEGIN PROLOGUE  LPDP
4C***REFER TO  LSEI
5C
6C     DIMENSION A(MDA,N+1),PRGOPT(*),X(N),WS((M+2)*(N+7)),IS(M+N+1),
7C     where N=N1+N2.  This is a slight overestimate for WS(*).
8C
9C     Written by R. J. Hanson and K. H. Haskell, Sandia Labs
10C     Revised Oct. 1, 1981.
11C
12C     Determine an N1-vector W, and
13C               an N2-vector Z
14C     which minimizes the Euclidean length of W
15C     subject to G*W+H*Z .GE. Y.
16C     This is the least projected distance problem, LPDP.
17C     The matrices G and H are of respective
18C     dimensions M by N1 and M by N2.
19C
20C     Called by subprogram LSI( ).
21C
22C     The matrix
23C                (G H Y)
24C
25C     occupies rows 1,...,M and cols 1,...,N1+N2+1 of A(*,*).
26C
27C     The solution (W) is returned in X(*).
28C                  (Z)
29C
30C     The value of MODE indicates the status of
31C     the computation after returning to the user.
32C
33C          MODE=1  The solution was successfully obtained.
34C
35C          MODE=2  The inequalities are inconsistent.
36C***ROUTINES CALLED  SCOPY,SDOT,SNRM2,SSCAL,WNNLS
37C***END PROLOGUE  LPDP
38
39
40