How to optimise mixed integer programming

[drkanda]

Dear all,
My model is involved in mixed integer programming.
For example:
When I vary customer demands (unit of items) allocated to factories in order to minimise total costs,
I set them as INTEGER(customer demands). However; when I compare the values resulting from Vensim with LINDO/LINGO, they are not equal.
I am not sure whether Vensim can solve the mixed integer programming like LINDO/LINGO do or not.

Pls advise me how to deal with this situation.

Thx in advance
drkanda

[tomfid]

When you say that the values are different, presumably you mean that Vensim identifies a different optimal solution. I don't know what your system is like, but this actually seems like a rather likely outcome in a discrete problem, unless you do a LOT of exploration of the payoff with random multistart optimization in Vensim. If you have a problem that really can be represented as an LP or MIP, you probably are better off with other tools.

Tom

[tomfid]

I neglected to mention that some allocation problems can be solved with the new integer-constrained functionality of the ALLOCATE AVAILABLE function.

I once succeeded in connecting lindo to Vensim via an external function, making LP solutions available. The code was never commercial quality, but it's a fairly straightforward task to repeat.

Tom

[drkanda]

Thx for your advice