[rules-users] Drools Planner: Linear programming with drools?

[Geoffrey De Smet]

Op 15-11-12 14:26, oliver.rolle schreef:
> Hi,
> I am writing a paper about route optimizations for fire trucks through
> influencing the traffic control system. The idea is to give fire trucks a
> green wave under the restriction of the flow of traffic.
Interesting.
>
> I want to compare the solutions of different methods for the problem:
> 1. Linear programming / Simplex: best solution
Simplex will give only give you the optimal solution if you have enough 
time to let it run that long.
For medium/large cases with a real number of constraints, this can 
easily be years.

In practice, it is never about finding the optimal solution,
it's about finding the best solution in a fixed amount of time.
> 2. Heuristic / Drools: good solution
> 3. Real-time / Drools: good solution
>
> The simplex algorithm is not implemented in Drools. Can Drools emulate the
> simplex algorithm?
No. Planner doesn't force you to use only linear constraints.
>   Does Drools metaheuristic approach solve linear models as
> good as simplex does? always?
In the experiments and competitions I 've been involved with [1], I 've 
seen the following:
- Simplex is better at scaling down (< 100 variables, < 100 values)
- Drools Planner is far better at scaling up (> 10 000 variables, > 10 
000 values) (Simplex goes out of memory)
- In the range between, they average out.
- Drools Planner can do real-time planning: good solutions in ms after 
the last problem change.

[1] Google Roadef 2012, ITC 2007, INRC 2010, ...
>
> best regards
> Oli
>
>
>
>
> --
> View this message in context: http://drools.46999.n3.nabble.com/Drools-Planner-Linear-programming-with-drools-tp4020844.html
> Sent from the Drools: User forum mailing list archive at Nabble.com.
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-users
>