[rules-users] OptaPlanner scalability

[Geoffrey De Smet]

In an experiment I did over 1 year ago (with an older version of 
optaplanner):
https://github.com/ge0ffrey/traveling-santa-problem/tree/master/src/main/java/org/droolsplannerdelirium/travelingsanta/solver/move
Warning: Here be dragons. That entire repo/code is a hack, an experiment 
and has no intention to live up the quality standards of the optaplanner 
code or examples...

But that experiment is not really nearby selection (it doesn't use a 
beta distribution IIRC).

I am hoping to start at nearby selection in optaplanner-core itself, in 
a user-friendly, powerful implementation, in a week or 2.

On 07-05-14 08:21, Rupesh M G wrote:
> Kindly give some more ideas on how you experimented the nearby selection.
> I would like to try that in my application.
>
> Regards,
> Rupesh
>
> -----Original Message-----
> From: rules-users-bounces at lists.jboss.org [mailto:rules-users-bounces at lists.jboss.org] On Behalf Of Geoffrey De Smet
> Sent: 30 April 2014 18:42
> To: rules-users at lists.jboss.org
> Subject: Re: [rules-users] OptaPlanner scalability
>
>
> On 30-04-14 15:04, Rupesh M G wrote:
>> Hi,
>>
>> I tested Optaplanner successfully for a fleet planning application.
>> Upto 1200 trips with around 80 vehicles for 3 days is fine.
>> I'm reaching the target of 0 hard constraints in less than 30 minutes.
>>
>> But my target is to plan for a month with 12000 trips using 80 vehicles.
>> This time it didn't complete even after 12 hours.
>> I found the memory (< 500 MB) and CPU (25%) usage is less.
>> So tried running 3 solver threads and grouping the vehicles.
>> Now CPU usage increases.
>>
>> I followed the vehicle routing sample application and the config xml is very similar.
>> I used late acceptance (200) and accept count (1000).
>>
>> I'm using incremental score calculation; and applied change & swap filters in the config xml.
>>
>> Requesting experts for some thoughts on how Optaplanner can scale on similar situations.
> See the 2 blog articles by Roman about VRP in this section:
>     http://www.optaplanner.org/learn/testimonialsAndCaseStudies.html
> He uses geo-fencing to scale.
>
> I don't believe geo-fencing is the best approach, I am working on nearby selection, which I believe will be better (because it still allows any location to go to any other location):
>     https://issues.jboss.org/browse/PLANNER-202
> Some time ago, I did an experiment with nearby selection on a TSP variant with 150 000 locations and it worked well.
>
> Also, the 6.1 beta's have seen some small perf improvements related to vehicle routing IIRC.
>
>> Thanks & Regards,
>> Rupesh
>>
>>
>>
>>
>>
>> DISCLAIMER: "The information in this e-mail and any attachment is intended only for the person to whom it is addressed and may contain confidential and/or privileged material. If you have received this e-mail in error, kindly contact the sender and destroy all copies of the original communication. IBS makes no warranty, express or implied, nor guarantees the accuracy, adequacy or completeness of the information contained in this email or any attachment and is not liable for any errors, defects, omissions, viruses or for resultant loss or damage, if any, direct or indirect."
>>
>> _______________________________________________
>> rules-users mailing list
>> rules-users at lists.jboss.org
>> https://lists.jboss.org/mailman/listinfo/rules-users
>>
>
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-users
>
>
> DISCLAIMER: "The information in this e-mail and any attachment is intended only for the person to whom it is addressed and may contain confidential and/or privileged material. If you have received this e-mail in error, kindly contact the sender and destroy all copies of the original communication. IBS makes no warranty, express or implied, nor guarantees the accuracy, adequacy or completeness of the information contained in this email or any attachment and is not liable for any errors, defects, omissions, viruses or for resultant loss or damage, if any, direct or indirect."
>
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-users
>