Page 1 of 1

### Unbounded Solution

Posted: Fri Nov 24, 2017 7:54 am
Hi All,
I am trying to decompose model(may be it is also called multi-objective problem ) in such a way a that value of variables estimated in first sub model can be used for second sub model. But in doing so second model leading to infeasible solution due to unbounded. Please give me suggestions how to formulate this problem so that variable optimize in sub problem 1 use information as a input for second model.
Below is code

*Model 1
positive variable f1,f2, x,y;
variable f3 ;
equation
a1,a2,a3;

a1.. f1 =e= 10*x+9*y ;
a2.. f2 =e= 9*x+8*y ;
a3.. f3 =e= f1-f2;

x.lo = 1;
y.lo = 1;
x.up = 5;
y.up = 3 ;
model m /a1,a2,a3/;
solve m maximizing f3 using lp ;
* Model 2
scalar IR /1.1/
variable f4 ;
equation a4;
a4.. f4 =e= f3/IR**IR;
model m1 /a4/;
solve m1 maximizing f4 using lp;

### Re: Unbounded Solution

Posted: Tue Nov 28, 2017 10:10 am
Hi Rofice

I don't understand why you are maximizing in the second model. If you have f3 from the first optimization, you just use it to calculate f4.
f4 = f3.L/IR**IR;

If you want to maximize F4 by letting F3 free (as you do in the second model) it is obvious, that this is unbounded, as there is no restriction on F3. If you want to keep this model, you should set F3 to F3.FX, but, then again, for that you don't need a model.
Cheers
Renger

### Re: Unbounded Solution

Posted: Tue Nov 28, 2017 12:07 pm
Dear Renger,
Thank you for your response, I caught your hint. Actually I was not aware that I can use variable f3 as f3.l and then define f4 as a parameter and find f4. Thanks for sharing knowledge.

Warm Regards,
Rofice