Re: [rules-users] Drools Planner: Vehicle routing problems
Hi Ge0ffrey, What do you mean by initial solution? The setPlanningProblem() aka the uninitialed solution or the solution after the construction heuristic (aka the initialized solution) I mean the initialised solution What's 109? Sorry, the number is just the number of milliseconds as returns by the new solution event Ok, if it's invalid here, it's invalid during the construction heuristic already. so at this point the solution is valid, its later steps where I get the problem At 127: D = Crew2 A = B = Crew1 B = A = B = Crew1 How is this possible? That means that B is pointing both to A and to Crew1 ? It's only a single pointer? VrpCustomer.getPreviousAppereance() Indeed this is my problem. There seems to be more than 1 instance of B. I've debugged my example code, and then issue seems to be: https://github.com/davidglassborow/drools-planner/blob/vehicle_routing/drools-planner-examples/src/main/java/org/drools/planner/examples/vehiclerouting/domain/Job.java#L33 The solver seems to be assigning the visit to be before itself ! If you look at: https://github.com/davidglassborow/drools-planner/blob/vehicle_routing/drools-planner-examples/src/main/java/org/drools/planner/examples/tsp/domain/Visit.java#L59 you'll see I have added a check to your Visit.java, which doesn't display the same problem. So I'm not sure what is happening, domain objects seems to be cloned and assigned to themselves. I have checked and double-checked the code inline with your TSP example, I'm sure it's nothing stupid I've done but can't for the life of me see anything wrong. If you download the branch, and run the VehicleRoutingHelloWorld class, you'll see the problem occuring. I hacked TSP to do multi TSP (see my google+ image), and it worked without a problem, so there must be a difference somewhere in your code. What did you need to change to get that to work ? Was it just a case of adding another domicile, and then removing the return to domicile cost rule ? Cheers for the help, Dave -- View this message in context: http://drools.46999.n3.nabble.com/Drools-Planner-Vehicle-routing-problems-tp3772797p3780139.html Sent from the Drools: User forum mailing list archive at Nabble.com. ___ rules-users mailing list rules-users@lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users
Re: [rules-users] Drools Planner: Vehicle routing problems
Geoffrey, Check your Solution's cloning method. If you clone the Visit, you need to make sure that planning variables of each cloned Visit are pointing to the new clones and not the original Visits. See my implementation in TSP. Sure, my clone method is virtually the same as your TSP one (my whole project was basically a copy of TSP): https://github.com/davidglassborow/drools-planner/blob/vehicle_routing/drools-planner-examples/src/main/java/org/drools/planner/examples/tsp/domain/TravelingSalesmanTour.java#L96 and https://github.com/davidglassborow/drools-planner/blob/vehicle_routing/drools-planner-examples/src/main/java/org/drools/planner/examples/vehiclerouting/domain/VehicleRouter.java#L41 Did you have to make any other changes to make multi TSP work ? Was it just a case of adding another domicile, and then removing the return to domicile cost rule ? Cheers, Dave -- View this message in context: http://drools.46999.n3.nabble.com/Drools-Planner-Vehicle-routing-problems-tp3772797p3780331.html Sent from the Drools: User forum mailing list archive at Nabble.com. ___ rules-users mailing list rules-users@lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users
[rules-users] Drools Planner: Vehicle routing problems
Hi, as discussed the other day with Ge0ffrey on IRC, I've been trying to get some vehicle routing done using the new TSP / chaining functionality in drools planner. I've written an example, but am having problems with the chaining of jobs together. After pulling my hair out, I've put together an example into my fork of planner, showing a simplified example of how I am approaching it, and the problem that can be replicated. https://github.com/davidglassborow/drools-planner/tree/vehicle_routing It is a very simple console app along the same lines as the hello world example for NQueens. On running it will print out the chains, and you will see the some Jobs appear more than once in the same list. On the planning variable it seems to be get assigned in front of itself, I've put some notes in the source, and checked that the same doesn't occur in your TSP example. I'm not sure if its a bug in the way I've setup things, or whether its an issue in the chain move code. Greatly appreciated if anybody in the know gets a chance to have a look. Thanks, David -- View this message in context: http://drools.46999.n3.nabble.com/Drools-Planner-Vehicle-routing-problems-tp3772797p3772797.html Sent from the Drools: User forum mailing list archive at Nabble.com. ___ rules-users mailing list rules-users@lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users
Re: [rules-users] Drools Planner: Vehicle routing problems
Some more details on my problem: Here is an example of a vehicle routing solution, 2 crews and 4 Jobs A, B, C and D. The print out below is the solver and solutions found at each step (it prints out each chain of jobs) As you can see the initial solution is valid. The next and last solutions are invalid though, with the same job appearing more than once in a chain. At 109: B = Crew1 D = Crew2 A = B = Crew1 C = A = B = Crew1 2012-02-24 15:20:39,160 [main] DEBUG Step index (3), time spend (110), score (0hard/-17soft), initialized planning entity (Job D ( linked to Crew Crew2)). 2012-02-24 15:20:39,160 [main] INFO Phase construction heuristic ended: step total (4), time spend (110), best score (0hard/-17soft). At 127: D = Crew2 A = B = Crew1 B = A = B = Crew1 C = A = B = Crew1 2012-02-24 15:20:39,177 [main] DEBUG Step index (0), time spend (127), score (0hard/-15soft), new best score (0hard/-15soft), accepted move size (18) for picked step (Job B ( linked to Crew Crew1) = Job A ( linked to Job B)). 2012-02-24 15:20:39,191 [main] DEBUG Step index (1), time spend (141), score (0hard/-15soft), best score (0hard/-15soft), accepted move size (14) for picked step (Job D ( linked to Crew Crew2) = Crew1). 2012-02-24 15:20:39,206 [main] DEBUG Step index (2), time spend (156), score (0hard/-16soft), best score (0hard/-15soft), accepted move size (10) for picked step (Job C ( linked to Job A) = Job B ( linked to Job A)). 2012-02-24 15:20:39,223 [main] DEBUG Step index (3), time spend (173), score (0hard/-16soft), best score (0hard/-15soft), accepted move size (5) for picked step (Job A ( linked to Job B) = Job B ( linked to Job A)). 2012-02-24 15:20:39,239 [main] WARN Cancelled step index (4), time spend (189): there is no doable move. Terminating phase early. 2012-02-24 15:20:39,239 [main] INFO Phase local search ended: step total (4), time spend (189), best score (0hard/-15soft). 2012-02-24 15:20:39,239 [main] INFO Solving ended: time spend (189), best score (0hard/-15soft), average calculate count per second (1089). Best: D = Crew2 A = B = Crew1 C = A = B = Crew1 B = A = B = B = Crew1 -- View this message in context: http://drools.46999.n3.nabble.com/Drools-Planner-Vehicle-routing-problems-tp3772797p3772841.html Sent from the Drools: User forum mailing list archive at Nabble.com. ___ rules-users mailing list rules-users@lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users
