Skip to content
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

Illustrating local pruning issue in original paper #63

Open
Catatomik opened this issue Oct 24, 2023 · 0 comments
Open

Illustrating local pruning issue in original paper #63

Catatomik opened this issue Oct 24, 2023 · 0 comments
Assignees
Labels
bug Something isn't working documentation Improvements or additions to documentation

Comments

@Catatomik
Copy link
Member

Catatomik commented Oct 24, 2023

local pruning issue in original paper

Problem

local pruning should work fine, but its implementation in Algorithm 1: RAPTOR seems bad so it doesn't work properly.
Indeed, line 18, the check is only done by comparing the arrival time with τ* and not t_k (which is expected) ; but τ* isn't updated when checking for foot-paths, τ_k is the only one assigned.
So back to the comparison line 18, paths found thanks to foot-paths aren't taken into account.
This eventually leads to cyclic paths, and produces wrong shortest paths.

Source : RAPTOR.

Solution

Remove local pruning improvment.

Illustrating

Branch no-local-pruning shows a version of RAPTOR without any local pruning upgrade, theoretically working.
Branches test-bibm and test-bibm-no-local-pruning implement tests based on BIBM data, demonstrating the issue.

@Catatomik Catatomik added bug Something isn't working documentation Improvements or additions to documentation labels Oct 24, 2023
@Catatomik Catatomik pinned this issue Oct 25, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working documentation Improvements or additions to documentation
Projects
None yet
Development

When branches are created from issues, their pull requests are automatically linked.

2 participants