[Coin-discuss] Controlling the level of output from solvers

[Michal Kaut]

Hello,

is there any way in Osi to completely switch off all the output from the 
solvers, or redirect somewhere? I solve a problem iteratively and my own 
messages get lost in all the output...

I tried "setHintParam(OsiDoReducePrint,true,OsiHintDo)", but it does not 
help much - for example OsiGlpkSolverInterface does not seem to 
implement setHintParam at all..

With OsiSolverGlpk, I even tried
- lpx_set_int_parm(GLPK_model, LPX_K_MSGLEV, 0);
- lpx_set_int_parm(GLPK_model, LPX_K_OUTFRQ, 1000);
but even this does not help. (I noticed that for ex. resolve() resets 
LPX_K_MSGLEV to 1 at start, so there is not point changing it anyway, is 
it?)
I still get the following output for each call of branchAndBound() 
(which I changed so that it uses lpx_intopt() instead of lpx_integer()):
------------------------------------------------------------------------
ipp_basic_tech:  41 row(s) and 104 column(s) removed
ipp_reduce_bnds: 3 pass(es) made, 23 bound(s) reduced
ipp_basic_tech:  2 row(s) and 0 column(s) removed
ipp_reduce_coef: 2 pass(es) made, 12 coefficient(s) reduced
lpx_intopt: presolved MIP has 37 rows, 88 columns, 200 non-zeros
lpx_intopt: 69 integer columns, all of which are binary
lpx_adv_basis: size of triangular part = 35
Solving LP relaxation...
     173:   objval =  0.000000000e+000   infeas =  1.000000000e+000 (2)
     183:   objval =  4.900000000e+001   infeas =  0.000000000e+000 (2)
*   183:   objval =  4.900000000e+001   infeas =  0.000000000e+000 (2)
*   197:   objval =  2.700000000e+001   infeas =  0.000000000e+000 (2)
OPTIMAL SOLUTION FOUND
Integer optimization begins...
+   197: mip =     not found yet >=              -inf        (1; 0)
+   200: mip =  3.150000000e+001 >=  2.700000000e+001  14.3% (2; 0)
+   202: mip =  3.150000000e+001 >=     tree is empty   0.0% (0; 3)
INTEGER OPTIMAL SOLUTION FOUND
------------------------------------------------------------------------

Any idea how to get rid of it?
If there is no general OsiSolverInterface solution, I would appreciate 
help for Glpk, Cbc and/or Cplex.

Thanks a lot in advance.


Regards,
Michal Kaut