[Help-glpk] API usage : explicit scaling versus presolving

[Robbie Morrison]

Hello Andrew and GLPK list

I would like to confirm the following behavior.

Does electing to use the simplex presolver, namely:

  glp_smcp parmSimplex;
  glp_init_smcp(&parmSimplex);
  parmSimplex.presolve = GLP_ON;
  ...
  int solverRet = glp_simplex(prob, &parmSimplex);

negate the effect of (and the resources invested in)
any previous scale call:

  glp_scale_prob(prob, GLP_SF_AUTO);

or basis call, for that matter:

  glp_adv_basis(prob, 0);

Moreover, is it okay that the different scalings give different
results (the logging from host application program is
also shown).

<terminal>

line   source           call                     no   delta-t    rank  
message
....................................................................................................

2735   d/siglp.cc       solverInvokeSolver     1465   00.0000s   dbug  
GLPK glp_scale_prob call

Scaling...
 A: min|aij| =  1.000e+00  max|aij| =  3.499e+06  ratio =  3.499e+06
GM: min|aij| =  5.419e-01  max|aij| =  1.845e+00  ratio =  3.406e+00
EQ: min|aij| =  3.030e-01  max|aij| =  1.000e+00  ratio =  3.300e+00

2762   d/siglp.cc       solverInvokeSolver     1466   00.0001s   adhc  
GLPK construct adv basis omitted
2818   d/siglp.cc       solverInvokeSolver     1467   00.0000s   dbug  
GLPK glp_simplex call

GLPK Simplex Optimizer, v4.44
13 rows, 9 columns, 23 non-zeros
Preprocessing...
4 rows, 4 columns, 9 non-zeros
Scaling...
 A: min|aij| =  1.000e+00  max|aij| =  3.499e+06  ratio =  3.499e+06
GM: min|aij| =  9.194e-01  max|aij| =  1.088e+00  ratio =  1.183e+00
EQ: min|aij| =  8.453e-01  max|aij| =  1.000e+00  ratio =  1.183e+00
Constructing initial basis...
Size of triangular part = 4
      0: obj =   5.132003000e+08  infeas =  1.778e+00 (0)
      2: obj =   5.132003001e+08  infeas =  3.678e-04 (0)
PROBLEM HAS NO FEASIBLE SOLUTION
glp_simplex: unable to recover undefined or non-optimal solution

2849   d/siglp.cc       solverInvokeSolver     1468   00.0003s   dbug   LP
relaxation FAIL, solver return 10

</terminal>

Clearly the nonfeasible problem is my mistake .. oops!

Robbie
---
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]