Model does not find the optimal solution - AOA Solver

[Florian Dulle]

I designed a modell and think all the constraints should be alright. (There still can be a mistake, because I am a newbie) AIMMS does find a solution, but I know it is not the best one. Is there an option to improve my model or the MainExecution so that AIMMS can find a better/globally optimal solution? I use the AOA Solver.

The Object is to minimize "Gesamtkosten" by determining the value of "Entscheidung(r,m)".

The solution for data "problem" is Gesamtkosten = 400. But it should be possible to get a better solution. For example:

Gesamtkosten = 145 and Monategesamt = 46

Entscheidung(r,m) is zero for all variables except:

Entscheidung(1,17)=1

Entscheidung(2,27)=1

Entscheidung(3,37)=1

Is 100 Iterations not enough?

I also tried to use GMP, but it does not work either.

I would really appreciate you help!

[Florian Dulle]

I could not figure out yet, what is wrong, but I got an idea:

In all the solutions I get the sum of Entscheidung(r,m) for r=0 equals 1 or more. But it there are solutions for that the sum should be zero and the sum of Entscheidung(r,m) for r=1 equals 1.

I cannot find a constraint, that makes the sum of Entscheidung(r,m) for r=0 1 or more, but I see a possible problem here.

[Florian Dulle]

I found the mistake. I used the index r with an elementrange(0, 5). Starting with "0" was the problem. elementrange(1, 6) does work now.

On Thursday, September 7, 2017 at 6:55:54 PM UTC+2, Florian Dulle wrote:

[Marcel Hunting]

Hi,

Your definition of the variable gesamtzunahme(t) is:

  if zunahme(t) = 0 then gesamtzunahme(t-1)

  else gesamtzunahme(t-1)+zunahme(t)

  endif;

This is equivalent to the linear definition:

   gesamtzunahme(t-1)+zunahme(t)

Namely, if zunahme(t) = 0 then gesamtzunahme(t-1)+zunahme(t) = gesamtzunahme(t-1).

In that case your model becomes linear.

Best regards,

Marcel Hunting

AIMMS Optimization Specialist