In this doctoral thesis, four questions related to computational geometry are considered. The first is an extremal combinatorics question regarding triangles with vertices taken from a set of n points in convex position. More precisely, two such triangles can exhibit eight distinct configurations and, for each subset of these configurations, we are interested in the asymptotics of how many triangles one can have while avoiding configurations in the subset (as a function of n). For most of these subsets, we answer this question optimally up to a logarithmic factor in the form of several Turán-type theorems. The answers for the remaining few were in turn tied to that of a long-standing open problem which appeared in the literature in the contexts of monotone matrices, tripod packing and 2-comparable sets.
The second problem, called Line Segment Covering (LSC), is about covering the cells of an arrangement of line segments with these line segments, where a segment covers the cells it is incident to. Recently, a PTAS, an APX -hardness proof and a FPT algorithm for variants of this problem have been shown. This paper and a
new slight generalization of one of its results is included as a chapter.
The third problem has been posed in the Sixth Annual Workshop on Geometry and Graphs and concerns the design of road networks to minimize the maximum travel time between two point sets in the plane. Traveling outside the roads costs more time per unit of distance than traveling on the roads and the total length of the roads can not exceed a budget. When the point sets are the opposing sides of a unit square and the budget is at most √2, we were able to come up with a few network designs that cover all possible cases and are provably optimal. Furthermore, when both point sets are the boundary of a unit circle, we managed to disprove the natural conjecture that a concentric circle is an optimal design.
Finally, we consider collision-avoiding schedules of unit-velocity axis-aligned trains departing and arriving from points in the integer lattice. We prove a few surprising results on the existence of constant upper bounds on the maximum delay that are independent of the train network. In particular, these upper bounds are shown to always exist in two dimensions and to exist in three dimensions for unit-length trains. We also showed computationally that, for several scenarios, these upper bounds are tight.
| Identifer | oai:union.ndltd.org:uottawa.ca/oai:ruor.uottawa.ca:10393/40569 |
| Date | 29 May 2020 |
| Creators | Schultz Xavier da Silveira, Luís Fernando |
| Contributors | Dujmovic, Vida, Bose, Prosenjit, Morin, Pat |
| Publisher | Université d'Ottawa / University of Ottawa |
| Source Sets | Université d’Ottawa |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Format | application/pdf |
Page generated in 0.0028 seconds