Originally posted by: dbaena

hi, i have a problem and i don't know what happens...i explain:

I have the same problem, I use Cplex C++ Library, cplex 12.1 If I use simplex primal I obtain:

Using devex.

Iteration log . . .
Iteration:     1    Objective     =             0.000000
Iteration:    21    Objective     =             0.000000
Iteration:   239    Objective     =             0.000000
Iteration:   384    Objective     =             0.000000
Iteration:   521    Objective     =             0.000000
Perturbation started.
Iteration:   602    Objective     =             0.000000
Iteration:   756    Objective     =             0.022364
Iteration:   897    Objective     =             0.039677
Iteration:  1035    Objective     =             0.053050
Iteration:  1176    Objective     =             0.064888
Iteration:  1319    Objective     =             0.077251
Iteration:  1452    Objective     =             0.087670
Iteration:  1589    Objective     =             0.098455
Iteration:  1723    Objective     =             0.107317
Iteration:  1856    Objective     =             0.116681
Iteration:  1990    Objective     =             0.127404
Iteration:  2141    Objective     =             0.138282
Iteration:  2294    Objective     =             0.147497
Iteration:  2436    Objective     =             0.155785
Iteration:  2583    Objective     =             0.161109
Iteration:  2735    Objective     =             0.165421
Iteration:  2888    Objective     =             0.170307
Iteration:  3050    Objective     =             0.175496
Iteration:  3209    Objective     =             0.181585
Iteration:  3360    Objective     =             0.186767
Iteration:  3523    Objective     =             0.193124
Iteration:  3693    Objective     =             0.200859
Iteration:  3852    Objective     =             0.207834
Iteration:  4028    Objective     =             0.215991
Iteration:  4207    Objective     =             0.223898
Iteration:  4393    Objective     =             0.231754
Iteration:  4596    Objective     =             0.240617
Removing perturbation.

 

If after,I do a little change in the objective function and in one constraint and I solve the problem using simplex primal, I obtain:

 

Using devex.

Iteration log . . .
Iteration:     1    Scaled infeas =             0.036395
Iteration:     8    Objective     =             0.000000
Iteration:   247    Objective     =             0.000000
Iteration:   262    Objective     =             0.000000
Iteration:   481    Objective     =             0.000000
Perturbation started.
Iteration:   610    Objective     =             0.000000
Iteration:   757    Objective     =             0.021236
Iteration:   909    Objective     =             0.040795
Iteration:  1065    Objective     =             0.053508
Iteration:  1217    Objective     =             0.064672
Iteration:  1375    Objective     =             0.074570
Iteration:  1523    Objective     =             0.086087
Iteration:  1675    Objective     =             0.097323
Iteration:  1832    Objective     =             0.108842
Iteration:  1989    Objective     =             0.120515
Iteration:  2173    Objective     =             0.136447
Iteration:  2355    Objective     =             0.157850
Iteration:  2509    Objective     =             0.181961
Removing perturbation.

 

 

So, it seems OK!!

 

Moreover, when I do getValues I obtain different values (maybe 0 or maybe not...).

 

However, if I use dual simplex...

The first time I solve this problem i obtain:

Initializing dual steep norms . . .

Iteration log . . .
Iteration:     1   Scaled dual infeas =             0.000000
Iteration:     2   Dual objective     =             1.000000
Perturbation started.
Iteration:   103   Dual objective     =             1.000000
Removing perturbation.
Iteration:   144   Dual objective     =             1.000000

 

However, when I do a little change in objective function and constraint (the same as before) I obtain:

 

Reinitializing dual norms . . .

 

and the same values that before!!!

 

So, someone knows what happens! It's very weird...

 

thanks a lot!!

 

 

 

 

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: BoJensen

Since one problem is your first model solved has different objectives for primal and dual, then try to write out the problem and load it into the interactive command prompt.  In the interactive command prompt see if you can reproduce the difference by running 'primopt' and 'tranopt'.

#CPLEXOptimizers
#DecisionOptimization