Christian Uhrig

The LEDA Platform of Combinatorial and Geometric Computing

We give an overview of the LEDA platform for combinatorial and geometric computing and an account of its development. We discuss our motivation for building LEDA and to what extent we have reached our goals. We also discuss some recent theoretical developments. This paper contains no new technical material. It is intended as a guide to existing publications about the system. We refer the reader also to our web-pages for more information.

Approximate motion planning and the complexity of the boundary of the union of simple geometric figures

We study rigid motions of a rectangle amidst polygonal obstacles. The best known algorithms for this problem have a running time of Ω(n2), wheren is the number of obstacle corners. We introduce thetightness of a motion-planning problem as a measure of the difficulty of a planning problem in an intuitive sense and describe an algorithm with a running time ofO((a/b · 1/ɛcrit + 1)n(logn)2), wherea ≥b are the lengths of the sides of a rectangle and ɛcrit is the tightness of the problem. We show further that the complexity (= number of vertices) of the boundary ofn bow ties (see Figure 1) isO(n). Similar results for the union of other simple geometric figures such as triangles and wedges are also presented.

Checking geometric programs or verification of geometric structures

A program checker verifies that a particular program execution is correct. We give simple and efficient program checkers for some basic geometric tasks. We report about our experiences with program checking in the context of the LEDA system. We discuss program checking for data structures that have to rely on user-provided functions.

Exact geometric computation in LEDA

Almost all geometric algorithms are based on the RealRAM model. Implementors often simply replace the exact real arithmetic of this model by fixed precision arithmetic, thereby making correct algorithms incorrect. Two approaches have been taken to remedy this sit uation. The first approach is redesigning geometric algorithms for fixed precision arithmetic. The redesign is difficult and can inherently not lead to exact results. Moreover it prevents application areas from making use of the rich literature of geometric algorithms developed in computational geometry. The other approach advocat es the use of exact real arithmetic.

Lower bounds for set intersection queries

We consider the followingset intersection reporting problem. We have a collection of initially empty sets and would like to process an intermixed sequence ofn updates (insertions into and deletions from individual sets) andq queries (reporting the intersection of two sets). We cast this problem in thearithmetic model of computation of Fredman [F1] and Yao [Ya2] and show that any algorithm that fits in this model must take time Ω(q+n√q) to process a sequence ofn updates andq queries, ignoring factors that are polynomial in logn. We also show that this bound is tight in this model of computation, again to within a polynomial in logn factor, improving upon a result of Yellin [Ye]. Furthermore, we consider the caseq=O(n) with an additional space restriction. We only allow the use ofm memory locations, wherem ≤n3/2. We show a tight bound of Θ(n2/m1/3) for a sequence ofn operations, again ignoring the polynomial in logn factors.

A lower bound for the worst case of Bottom-Up-Heapsort

Abstract Bottom-Up-Heapsort is a variant of Heapsort. Until now, its worst case complexity for the number of comparisons has been known to be bounded above by 1.5n log n + 0(n), where n is the number of elements to be sorted; but it was conjectured to be n log n + o(n log n). In this paper we give a construction that proves an asymptotic lower bound of 1.25n log n − 0(n log log n) comparisons for the worst case.

A computational basis for higher-dimensional computational geometry and applications

In this paper we describe and discuss a kernel for higher-dimensional computational geometry and we present its application in the calculation of convex hulls and Delaunay triangulations. The kernel is available in form of a software library module programmed in C++ extending LEDA. We introduce the basic data types like points, vectors, directions, hyperplanes, segments, rays, lines, spheres, affine transformations, and operations connecting these types. The description consists of a motivation for the basic class layout as well as topics like layered software design, runtime correctness via checking routines and documentation issues. Finally we shortly describe the usage of the kernel in the application domain.

Hidden line elimination for isooriented rectangles

We consider a special case of the hidden line elimination problem. The scene consists of n isooriented rectangles in R^{3} and an observer at z=+\infty. We give an algorithm that computes the visible parts of the edges on a random access machine in time O(n\log n+k\log\frac{n^{2}}{k}), where k is the number of line segments in the output, and uses O(nlog n) space.

k versus k + 1 index registers and modifiable versus non-modifiable programs

We compare Random Access Machines with k or k+1 index registers and modifiable or non-modifiable programs and show for a simple problem of data transfer that the more prowerful versions are more efficient.

Selected Topics from Computational Geometry, Data Structures and Motion Planning

The Voronoi diagram of a set of sites in the plane partitions the plane into regions, called Voronoi regions, one to a site. The Voronoi region of a site s is the set of points in the plane for which 8 is the closest site among all the sites. The Voronoi diagram has many applications in diverse fields, cf. Leven and Sharir [LS86] or Aurenhammer [Aur90] for a list of applications and a history of Voronoi diagrams. Different types of diagrams result from considering different notions of distance, e.g. Euclidean or Lp-norm or convex distance functions, and different sorts of sites, e.g. points, line segments, or circles. For many types of diagrams efficient construction algorithms have been found; these are either based on the divide-and-conquer technique due to Shamos and Hoey [SH75], the sweepline technique due to Fortune [For87], or geometric transforms due to Brown [Bro79] and Edelsbrunner and Seidel [ES86]. A unifying approach to Voronoi diagrams was proposed by Klein [Kle88a, Kle88b, Kle89a, KleS9b], cf. [ES86] for a related approach. Klein does not use the concept of distance as the basic notion but rather the concept of bisecting curve, i.e. he assumes for each pair {p, q} of sites the existence of a bisector J(p, q) which is homeomorphic to a line and divides the plane into a p-region and a q-region. The intersection of all p-regions for different qs is then the Voronoi-region of site p. He also postulates that Voronoi-regions are simply-connected and partition the plane. He shows that these so-called abstract Voronoi diagrams have already many of the properties of concrete Voronoi diagrams. In [MMO91] and the refinement [KMM91] we present a randomized incremental algorithm that can handle abstract Voronoi diagrams in (almost) their full generality. When n denotes the number of sites, the algorithm runs in O(nlog n) expected time, the average being taken over all permutations of the input. The algorithm is simple enough to be of great practical importance. It is uniform in the sense that only a single operation, namely the construction of a Voronoi diagram for 5 sites, depends on the specific type of Voronoi diagram and has to be newly programmed in order to adapt the algorithm to the type of the diagram. Moreover, this operation is the only geometric operation in our algorithm, and using this operation, abstract Voronoi diagrams can be constructed in a