Re: [Help-glpk] Perl and GnuLP kit

[Gabor Retvari]

On Thursday 10 August 2006 15:30, padma priya wrote:
> Hi
> I need info on how to use GnuLP kit with perl code. I
> want to solve a LP algorithm with Perl code. Can u pls
> send across details with sample code if possible on
> how to use th GnuLP kit.
> Thanks in advance
> Padma Priya v

Hello there,

I wrote the Math::GLPK module to facilitate doing linear programming by GLPK 
from within Perl scripts. The module and all its dependencies are available 
from here:

http://opt.tmit.bme.hu/~retvari/Math-GLPK.html

(note that this is a temporary place, since the main project page

http://qosip.tmit.bme.hu/~retvari/Math-GLPK.html

seems to be down currently).

Installing this module requires installing a recent version of 
Math::MatrixSparse and Math::GLPK::Solve, all available from 

http://opt.tmit.bme.hu/~retvari/

Then, solving LPs is as easy as:

#!/usr/bin/perl

# Solves a simple dual minimum cost multicommodity flow problem 
#############################################################
#
#  Sample problem: x is a column-vector of unknowns
#
#  max [-1 0 0 1 0 -1 0 1 0 0 0 0 0 0 0 0 -1 -1 -1 -1] * x
#
#  s.t.
#
#  [-1 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 -1 -0 -0 -0]       [ 1 ]
#  [0 -1 1 0 0 0 0 0 0 1 0 0 0 0 0 0 -0 -1 -0 -0]       [ 1 ]
#  [0 -1 0 1 0 0 0 0 0 0 1 0 0 0 0 0 -0 -0 -1 -0]       [ 1 ]
#  [0 0 -1 1 0 0 0 0 0 0 0 1 0 0 0 0 -0 -0 -0 -1] * x = [ 1 ]
#  [0 0 0 0 -1 1 0 0 0 0 0 0 1 0 0 0 -1 -0 -0 -0]       [ 1 ]
#  [0 0 0 0 0 -1 1 0 0 0 0 0 0 1 0 0 -0 -1 -0 -0]       [ 1 ]
#  [0 0 0 0 0 -1 0 1 0 0 0 0 0 0 1 0 -0 -0 -1 -0]       [ 1 ]
#  [0 0 0 0 0 0 -1 1 0 0 0 0 0 0 0 1 -0 -0 -0 -1]       [ 1 ]
#
#       x >= 0

use strict;
use warnings;

use Math::MatrixSparse;
use Math::GLPK qw(:constants);

# create the problem
# it has 8 constraints (rows) and 20 variables (columns)
# it is of the form
# max cx: Ax = b, x >= 0
print "Testing Math::GLPK by solving sample problem...\n";

my $lp = Math::GLPK->new(8, 20);

# set the name
$lp->prob_name("sample problem");

# set the direction of the optimization to MAX
$lp->obj_dir($LPX_MAX);

# set all variables to be non-negative
$lp->set_all_var_bnds($LPX_LO, 0);

# defined and add the objective (row vector)
$lp->objective( Math::MatrixSparse->new_from_rows([
 [ qw( -1 0 0 1 0 -1 0 1 0 0 0 0 0 0 0 0 -1 -1 -1 -1) ]
 ]) );

# define and add the right-hand side (column vector)
$lp->rhs( Math::MatrixSparse->new_from_cols([
 [qw(1 1 1 1 1 1 1 1)]
 ]) );

# define and the constraint matrix
$lp->constraint( Math::MatrixSparse->new_from_rows([
 [qw(-1 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 -1 -0 -0 -0)],
 [qw(0 -1 1 0 0 0 0 0 0 1 0 0 0 0 0 0 -0 -1 -0 -0)],
 [qw(0 -1 0 1 0 0 0 0 0 0 1 0 0 0 0 0 -0 -0 -1 -0)],
 [qw(0 0 -1 1 0 0 0 0 0 0 0 1 0 0 0 0 -0 -0 -0 -1)],
 [qw(0 0 0 0 -1 1 0 0 0 0 0 0 1 0 0 0 -1 -0 -0 -0)],
 [qw(0 0 0 0 0 -1 1 0 0 0 0 0 0 1 0 0 -0 -1 -0 -0)],
 [qw(0 0 0 0 0 -1 0 1 0 0 0 0 0 0 1 0 -0 -0 -1 -0)],
 [qw(0 0 0 0 0 0 -1 1 0 0 0 0 0 0 0 1 -0 -0 -0 -1)]
 ]) );

# use the presolver
$lp->int_parm($LPX_K_PRESOL, 1);

# apply scaling, before solving the problem
$lp->int_parm($LPX_K_SCALE, 1);

# set the debug level to maximum
$lp->int_parm($LPX_K_MSGLEV, 3);

# solve the problem by the simplex method
$lp->simplex;

# check if optimal solution is obtained
$lp->is_optimal or die "Could not obtain optimal solution!\n";

# print the results
print "Optimal objective is: " . $lp->get_obj_val . "\n";
print "Optimal solution is:\n";
$lp->get_opt_sol->pretty_print;

exit 0;

etc., see the test.pl file in the distribution.

If you have any problems or questions, feel free to contact me.

best regards,
gabor

p.s.: Math::GLPK currently only supports GLPK version <= 4.8 out of the box. I 
plan to release an update pretty soon...