[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Help-glpk] Semi-intelligent C++ wrapper for GLPK APIs

From: Robbie Morrison
Subject: [Help-glpk] Semi-intelligent C++ wrapper for GLPK APIs
Date: Sat, 3 Dec 2005 04:31:03 +1300 (NZDT)
User-agent: SquirrelMail/1.4.3a

Hello GLPK list

This posting should be of interest for those working directly
with the GLPK APIs.

An application programming interface (API) typically defines
the interface between client code and a given library.

The GLPK APIs are relatively low level in the sense that the
client code needs to know whether the current problem is LP
or MILP, which solver or solvers to call, how to interpret
return codes, and so forth.

I have written a C++ wrapper class which manages some of
this complexity from within that class.

The interface class is named 'SolverInterface_GLPK' and the
source files are named 'SiGlpk.*'.  A unit test file is also
available as 'SiGlpk.UnitTest.C'.  The code is licenced
under GNU GPL and thus fully compatible with GLPK.

At present, the class contains some functionality specific
to the application code I work with.  But writing out this
localization should be quite straightforward.  And
conversion to Java would also be relatively easy, I should

I am happy to email the source to anyone who is interested
(in about two weeks time).  If there is a certain level of
traffic, then I plan to release a generic version in about
six months time (my PhD write-up comes first otherwise).

A test version of the GLPK tutorial problem (page 9 of the
GLPK API manual) is given below.  This should give some
clues to the features provided by this interface class.

Many thanks, as always, to Andrew Makhorin and the wider
GLPK community.

regards to all

// -------------------------------------
//   GLPK API tutorial problem, page 9
// -------------------------------------

#include   <iostream>
#include     <string>

#include     "SiGlpk.h"  // GLPK solver interface class

int main(int argc, char* argv[])
// Specify problem meta-information

  int constraintCount = 3;  // total number of constraints
  int variableCount   = 3;  // total number of variables
  int coeffsEstimate  = 4;  // underestimate, correct value is 9

// Create solver interface object

  SolverInterface_GLPK* pSI = new SolverInterface_GLPK(

  // uncomment to use the optional simplex presolver
  // pSI->initUseSimplexPresolver();  // sets LPX_K_PRESOL

  // uncomment to use the interior point solver where appropriate
  // pSI->initSetPreferredLPSolver(si::solver_interior);

// Build problem

  // set objective sense

  // cols as non-zero objective coefficients
  pSI->loadObj(1, 10.0);
  pSI->loadObj(2,  6.0);
  pSI->loadObj(3,  4.0);

  // rows as non-negative RHS constraints
  pSI->loadRhs(1, 100, si::L);  // L indicates '<= rhs'
  pSI->loadRhs(2, 600, si::L);
  pSI->loadRhs(3, 300, si::L);

  // rows x cols as non-zero structural coefficients
  pSI->loadCoeff(1, 1,  1.0);
  pSI->loadCoeff(1, 2,  1.0);
  pSI->loadCoeff(1, 3,  1.0);

  pSI->loadCoeff(2, 1, 10.0);
  pSI->loadCoeff(2, 2,  4.0);
  pSI->loadCoeff(2, 3,  5.0);

  pSI->loadCoeff(3, 1,  2.0);
  pSI->loadCoeff(3, 2,  2.0);
  pSI->loadCoeff(3, 3,  6.0);

  // uncomment to convert to an MILP problem
  // pSI->markVarInteger(2);  // mark col 2 as integer

// Call solver

  pSI->runSolver();  // reports to console as well

// Print calculated results

  int vars = pSI->getVarCount();
  std::cout << "is LP : "
            << pSI->telIsLP()
            << std::endl;
  std::cout << "Z     : "
            << pSI->getObjValue()
            << std::endl;
  for (int var = 1; var <= vars; var++)
      std::cout << "x"
                << var
                << "    : "
                << pSI->getVarValue(var)
                << std::endl;

// Print known results

  std::cout << "KNOWN LP RESULTS" << "\n"
            << "Z  = 733.333"     << "\n"
            << "x1 = 33.3333"     << "\n"
            << "x2 = 66.6667"     << "\n"
            << "x3 = 0"           << "\n"
            << std::endl;

// Housekeeping

  delete pSI;  // free memory
  return 0;    // return from main()

}  // code ends

Robbie Morrison
PhD student -- policy-oriented energy system simulation
Institute for Energy Engineering (IET)
Technical University of Berlin (TU-Berlin), Germany
University email (redirected) : address@hidden
Webmail (preferred)           : address@hidden
[from Webmail client]

reply via email to

[Prev in Thread] Current Thread [Next in Thread]