Originally posted by: SystemAdmin

I am using CPLEX 12.3 on Windows XP to solve a set of LPs.
The attached file (myprob_0.lp) is a LP model whose coefficients c2 and c6 are zeros. I use CPXchgobj to change the coefficients. Below is the code that is used for this purpose:

double   *obj_value = 

new 

double[5];
// Array of the new objective coefficients 

int      *obj_index = 

new 

int[5];   
// Array of the indices of the new objective coefficients   

for (j=0; j < 5; j++) 
{ obj_value[j] = k(j); obj_index[j] = j; 
}   status = CPXchgobj (env, lp, 5, obj_index, obj_value); 

if (status) 
{ fprintf(stderr, 
"Failed to change the objective coefficients.\n", status); system(
"pause"); 
}   status = CPXwriteprob (env, lp, 
"myprob_1.lp", NULL); 

if (status) 
{ fprintf(stderr, 
"CPXwriteprob failed to write the LP Problem.\n", status); system(
"pause"); 
}

The array k of the new coefficients cj is (1.12 0.87 176 5 3.6). After applying the change, the new objective function is obj: 1.12 u1 + 0.87 u3 + 176 u4 + 5 u5 + 3.6 v1
The right coefficients are assigned to the wrong variables. This should be: obj: 1.12 u1 + 0.87 u2 + 176 u3 + 5 u4 + 3.6 u5
Is there any possible reason for this?
I appreciate your help.

Amar
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: T_O

Do you create your problem by reading data from the lp-file? Then the order of the variables in the problem is (if I remember correctly) the same order in which the variables appear in the lp-file ( u1, u3, u4, u5, v1, ... ). So there is nothing wrong in your code, you just misinterpret some things.

To avoid such problems, you could use mps or sav-files or set the objective coefficients by name.

Best regards,
Thomas
#CPLEXOptimizers
#DecisionOptimization