We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.OC

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Data Structures and Algorithms

Title: Scalable Algorithms for Bicriterion Trip-Based Transit Routing

Abstract: This paper proposes multiple extensions to the popular bicriterion transit routing approach -- Trip-Based Transit Routing (TBTR). Specifically, building on the premise of the HypRAPTOR algorithm, we first extend TBTR to its partitioning variant -- HypTBTR. However, the improvement in query times of HyTBTR over TBTR comes at the cost of increased preprocessing. To counter this issue, two new techniques are proposed -- a One-To-Many variant of TBTR and multilevel partitioning. Our One-To-Many algorithm can rapidly solve profile queries, which not only reduces the preprocessing time for HypTBTR, but can also aid other popular approaches such as HypRAPTOR. Next, we integrate a multilevel graph partitioning paradigm in HypTBTR and HypRAPTOR to reduce the fill-in computations. The efficacy of the proposed algorithms is extensively tested on real-world large-scale datasets. Additional analysis studying the effect of hypergraph partitioning tools (hMETIS, KaHyPar, and an integer program) along with different weighting schemes is also presented.
Subjects: Data Structures and Algorithms (cs.DS); Optimization and Control (math.OC)
Cite as: arXiv:2111.06654 [cs.DS]
  (or arXiv:2111.06654v2 [cs.DS] for this version)

Submission history

From: Tarun Rambha [view email]
[v1] Fri, 12 Nov 2021 10:46:21 GMT (937kb,D)
[v2] Sat, 26 Feb 2022 16:41:32 GMT (1146kb,D)

Link back to: arXiv, form interface, contact.