Originally posted by: Mehrsa

Hi every one

I have started learning opl and cplex on my own, I solve 2 simple problems but on the second one, The cplex gives me the solution that it does nt satisfy the constraints. I do nt know what I am doing wrong, please help me.

 

int m =...;

int n =...;

int s[1..n]=...;

int d[1..n]=...;

float t[1..m][1..n]=...;

float c[1..n]=...;

 

dvar int+ x [1..n];

 

maximize

sum (j in 1..n)( x[j]+s[j]-d[j]);

 

subject to {

forall (i in 1..m)

sum (j in 1..n)

t[i][j] * x[j]<=c[i];

forall (j in 1..n)

d[j]-s[j]<= x[j] ;

}

m=2;

n=2;

s=[30 90];

d=[75 95];

t=[[50 24]

[30 33]

];

c=[2400 2100]

 

the solution is (45,6) which is not correct.

 

 

#DecisionOptimization
#OPLusingCPLEXOptimizer

Hi

if you add

assert

 

forall (i in 1..m)

 

sum (j in 1..n)

 

t[i][j] * x[j]<=c[i];

 

assert

forall (j in 1..n)

 

d[j]-s[j]<= x[j] ;

in the postprocessing it works fine.

Which constraint is violated according to you ?

regards

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: Mehrsa

Thank you

I added assert before every forall as you said, it gives me an error ( an unexpected assert).

the first constraint is violated and in the problem browser it says solution with objective 1.

I just have 1 objective function and it doesn't give me any quantity with this solution.

#DecisionOptimization
#OPLusingCPLEXOptimizer

Hi,

you get the same solution if you write

dvar int x;

dvar int y;

 

maximize x+30-75+y+90-95;

subject to

{

50*x+24*y<=2400;

30*x+33*y<=2100;

x>=45;

y>=5;

}

Let me give you the model with asserts:

int m=...;

 

int n=...;

 

int s[1..n]=...;

 

int d[1..n]=...;

 

float t[1..m][1..n]=...;

 

float c[1..n]=...;

 

 

 

dvar int+ x [1..n];

 

 

 

maximize

 

sum (j in 1..n)( x[j]+s[j]-d[j]);

 

 

 

subject to{

 

forall (i in 1..m)

 

ct1:sum (j in 1..n)

 

t[i][j] * x[j]<=c[i];

 

forall (j in 1..n)

 

ct2:d[j]-s[j]<= x[j] ;

 

}

 

　

　

assert

 

forall (i in 1..m)

 

sum (j in 1..n)

 

t[i][j] * x[j]<=c[i];

 

assert

forall (j in 1..n)

 

d[j]-s[j]<= x[j] ;

 

 

To get some displey you may add

execute

{

writeln();

 

writeln("x=",x);

 

　

}

and you will get

 

x= [45 6]

regards

 

 

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: Mehrsa

Yes, the model I wrote is the same as yours, Although x=[45,6] is in feasible region, it is not the optimum, and that is my problem. because I solved it with simplex  in the paper.

#DecisionOptimization
#OPLusingCPLEXOptimizer

Hi,

so what would be the optimum ?

regards

#DecisionOptimization
#OPLusingCPLEXOptimizer