CODEWITHSUNDEEP
javabinary-datalinear-programmingglpk
sthor69
sthor69

Reputation: 668

GLPK for Java - Binary variable MIP gives fractional result

I'm trying to solve a Linear Programming problem based on binary variables with GLPK for Java (http://glpk-java.sourceforge.net/), but the result of the computation gives fractional results for the variables.

I omit most part of the code, but the important part is the following, where I define the variables as binary

GLPK.glp_add_cols(lp, data.size());
for (int i = 0; i < data.size(); i++) {
            GLPK.glp_set_col_name(lp, i + 1, "x" + (i + 1));
            GLPK.glp_set_col_kind(lp, i + 1, GLPKConstants.GLP_BV);
            }

Data is a table containing the coefficients.

If I try to solve the problem using the presolver

glp_iocp iocpParm = new glp_iocp();
iocpParm.setPresolve(GLPK.GLP_ON);
GLPK.glp_init_iocp(iocpParm);
ret = GLPK.glp_intopt(lp, iocpParm);

the result is an error

glp_intopt: optimal basis to initial LP relaxation not provided
The   problem   could  not  be  solved

If I add a pre-processing using simplex (as suggested by the documentation)

glp_smcp smcpParm = new glp_smcp();
GLPK.glp_init_smcp(smcpParm);
GLPK.glp_simplex(lp, smcpParm);

The results are fractional

Problem   created 
GLPK Simplex Optimizer, v4.63
1 row, 4 columns, 4 non-zeros
      0: obj =   0.000000000e+00 inf =   1.231e+03 (1)
      1: obj =   1.231000000e+03 inf =   0.000e+00 (0)
OPTIMAL LP SOLUTION FOUND
GLPK Integer Optimizer, v4.63
1 row, 4 columns, 4 non-zeros
4 integer variables, all of which are binary
Integer optimization begins...
+     1: mip =     not found yet >=              -inf        (1; 0)
Solution found by heuristic: 1600
+     2: >>>>>   1.400000000e+03 >=   1.400000000e+03   0.0% (1; 0)
+     2: mip =   1.400000000e+03 >=     tree is empty   0.0% (0; 1)
INTEGER OPTIMAL SOLUTION FOUND
z = 1231.0
x1 = 0.769375
x2 = 0.0
x3 = 0.0
x4 = 0.0

How can I get binary solutions?

Upvotes: 1

Views: 511

Answers (1)

user555045
user555045

Reputation: 64913

GLPK keeps separate results for different kinds of solver (MIP, interior point, simplex), so to get the results of a specific solver, the corresponding functions must be used.

  • To get results from using the simplex solver, use glp_get functions.
  • For the interior pointer solver, use glp_ipt functions.
  • For the MIP solver, use glp_mip functions.

The rest of the name of the function is a bit inconsistent.

Upvotes: 1

Related Questions