Messages

Hello Liviu,

thanks for the quick explanation. May I ask for better understanding more about the slow part?
- So as I understand it, it loops over all constraints so it has at least the complexity of O(n_constraints), right?
- Does evaluating one single constraint also involve looping (e.g. does evaluating constraint x involve looping over all activities / a subset of all activities, which will make evaluating x of the complexity O(n_activities))?
- Does evaluating some single constraints also involve nested loops?

Hallo Herr Dirr,

vielen Dank für die Hinweise. Tatsächlich habe ich darüber nachgedacht und würde das gerne genauer mit Ihnen besprechen. Könnten Sie sich eventuell vorstellen, über meine E-Mail Kontakt mit mir aufzunehmen?

Hi there,

I'm investigating FET at the moment and compare it to local search algorithms and wonder about the speed in terms of "steps".

Consider a "step" to be something like the following:
- evaluate, that activity A_1 can be put to timeslot T_1 and put it there.
- if no timeslot could be found and the activities B_1 to B_N have to be removed in order to place A_1, I consider this to be N steps.
- feel free to provide a better definition of an algorithm step.

I don't understand the algorithm good enough to know if there is even a meaningful definition of the term "step" for the FET algorithm. But as I don't need precise numbers but only a rough estimate, the definition doesn't need to be perfect.

Do you know any figures of how many steps FET performs per second roughly (say on an average environment)?

Furthermore: Do you know what is the slowest part of the algorithm during a step.
Parts may be:
- Evaluating if A_x in T_x breaks constraints.
- Looping inside the algorithm logic to decide the next step.