-
Notifications
You must be signed in to change notification settings - Fork 124
Prunning research #31
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
base: master
Are you sure you want to change the base?
Conversation
Hello, I looked at the code and wondered whether the new pruning rule makes a difference in performance. This is without your pruning rule:
and this is with:
A similar picture emerges with the changes to RAPTOR. If the number of outgoing transfers is rather low, then there is no point in not scanning the few footpaths. It could help if you have many footpaths per stop; I have not tested this. Without pruning rule:
and with:
|
Dear @PatrickSteil, Big thank you for your message, and I sincerely apologise for the unprepared PR. I made it coincidentally during my research. I was exploring the possibility of pruning the loop in the iterative footpath search part of algorithms that are not based on graphs, such as CSA and RAPTOR. I got the same results as you for CSA and did not see any performance improvement. However, there was a speed boost for RATPTOR and ULTRA-RAPTOR on the London and Switzerland datasets, which you shared publicly. The likely reason is that London and Switzerland are larger than Karlsruhe, and as you mentioned, there might be more transfer connections in the graph, which could be a factor. So, in larger cities, we might see speed-ups (+~25% for RAPTOR) (+~15% for ULTRA-RAPTOR) for these algorithms. Here are the results:
Also, here is the results for:
-ULTRA-RAPTOR (London)
P.S. I'll also add my code updates in this PR. I made some slight optimisations. Please let me know if you observe the same improvements on larger datasets. Big thank you for your quick response and validation of my idea, I really appreciate it! |
275a64c
to
445a418
Compare
Thanks heaps, @PatrickSteil, for sharing! I'll review it. Is it alright if I add this RAPTOR and ULTRA RAPTOR modification to your PR? |
Thanks for the offer! I’d prefer to keep them separate so Jonas and I can more easily distinguish what each branch does and who contributed which changes. |
Sure, thank you! Yeah, you are right. I will then review your modification and attempt to combine the ideas while keeping them separate. Big thank you for your help and suggestions! |
No description provided.