program test
*
* File:    ~/numerical/test_nag.f
*
* Author:  Chuck Gartland (6-99)
*
* Fortran test program for using NAG Library on the HP-UX 10.*,
* Linux, and SunOS 4.1 machines on the Kent Math/CS Network.
*
* Library locations:
*
*   HP-UX:       /local/opt/fortran/lib/libnag.a
*   Linux, Sun:  /usr/local/lib/libnag.a
*
* Compiling alternatives:
*
*   HP-UX:
*   -----
*     1) directing the loader to search extra directory
*
*       % fort77 test_nag.f -L/local/opt/fortran/lib -lnag
*
*     2) using full path name or link to the library archive
*
*       % f77 test_nag.f /local/opt/fortran/lib/libnag.a
*
*        or
*
*       % ln -s /local/opt/fortran/lib/libnag.a nag.a
*       % f77 test_nag.f nag.a
*
*     See man pages for "f77/fort77", "ld", and "ln".
*
*   Linux, Sun:
*   ----------
*     % f77 test_nag.f -lnag
*
* Test prob: [ 1 2 3 ][x]   [ 6] ,  w/ true sol'n  [1]
*            [ 4 5 6 ][y] = [15]                   [1]
*            [ 7 8 0 ][z]   [15]                   [1]
*
      implicit double precision ( a-h, o-z )

      dimension A(4,5), b(3), wkspce(3)         ! deliberately chosen larger 
                                                ! dimensions to illustrate role
                                                ! of 'ia'

      data A / 1., 4., 7., 0., 2., 5., 8., 0., 3., 6., 0., 0., 8*0. /
      data b / 6., 15., 15. /
      data ia, n, ifail / 4, 3, 0 /

      call F04ARF( A, ia, b, n, b, wkspce, ifail )          ! solve Ax=b system

      print *, ' ifail, b = ', ifail, b

      stop
      end
*
* Appended sample output:
*
*  ifail, b =  0 .9999999999999996 1.0 1.0
*