A linear-time algorithm for linearL1 approximation of points

In this paper we present a linear-time algorithm for approximating a set ofn points by a linear function, or a line, that minimizes theL ₁ norm. The algorithmic complexity of this problem appears not to have been investigated, although anO(n ³) naive algorithm can be easily obtained based on some simple characteristics of an optimumL ₁ solution. Our linear-time algorithm is optimal within a constant factor and enables us to use linearL ₁ approximation of many points in practice. The complexity ofL ₁ linear approximation of a piecewise linear function is also touched upon.     

The convergence rate of the sandwich algorithm for approximating convex functions

The Sandwich algorithm approximates a convex function of one variable over an interval by evaluating the function and its derivative at a sequence of points. The connection of the obtained points is a piecewise linear upper approximation, and the tangents yield a piecewise linear lower approximation. Similarly, a planar convex figure can be approximated by convex polygons. Different versions of the Sandwich algorithm use different rules for selecting the next evaluation point. We consider four natural rules (interval bisection, slope bisection, maximum error rule, and chord rule) and show that the global approximation error withn evaluation points decreases by the order ofO(1/n ²), which is optimal. By special examples we show that the actual performance of the four rules can be very different from each other, and we report computational experiments which compare the performance of the rules for particular functions.     

Approximate minimum weight matching on points ink-dimensional space

We study the problem of finding a minimum weight complete matching in the complete graph on a set V ofn points ink-dimensional space. The points are the vertices of the graph and the weight of an edge between any two points is the distance between the points under someL _q,-metric. We give anO((2c _q )^1.5k ?^?1.5k (α(n, n))^0.5 n ^1.5(logn)^2.5) algorithm for finding an almost minimum weight complete matching in such a graph, wherec _q =6k ^1/q for theL _q -metric, α is the inverse Ackermann function, and ? ≤ 1. The weight of the complete matching obtained by our algorithm is guaranteed to be at most (1 + ?) times the weight of a minimum weight complete matching.     

L1PMA: A Fortran 77 package for best L1 piecewise monotonic data smoothing

Fortran 77 software is presented for the calculation of a best L₁ approximation to n measurements that include random errors by requiring k−1 sign changes in the first divided differences of the approximation or equivalently k monotonic sections, alternately increasing and decreasing. A dynamic programming algorithm separates the measurements into optimal disjoint sections of adjacent data and applies to each section a single L₁ monotonic calculation. The most distinctive feature of the algorithm is that it terminates at a global minimum in at most n³+O(kn²) computer operations, although this calculation can exhibit O(n^k) local minima, because the optimal positions of the turning points are also unknowns of the optimization process. The arithmetic operations involved in this calculation are comparisons mainly spent in finding the medians of subranges of data during the monotonic calculations. The package employs techniques for median and for best L₁ monotonic approximation, while full details of these techniques are specified. The package has been applied and tested on a variety of data that have substantial differences and showed quadratic behaviour in n. Some numerical results demonstrate the performance of the method. Further, there is a commentary on the division of the code into subroutines. Driver programs and numerical examples with output are provided to help new users of the method. Besides that piecewise monotonicity is a property of a wide range of functions, an important application of the method is in estimating turning points of a function from some noisy measurements of its values.     

Optimal algorithms for some intersection radius problems

The intersection radius of a set ofn geometrical objects in ad-dimensional Euclidean space,E ^d , is the radius of the smallest closed hypersphere that intersects all the objects of the set. In this paper, we describe optimal algorithms for some intersection radius problems. We first present a linear-time algorithm to determine the smallest closed hypersphere that intersects a set of hyperplanes inE ^d , assumingd to be a fixed parameter. This is done by reducing the problem to a linear programming problem in a (d+1)-dimensional space, involving 2n linear constraints. We also show how the prune-and-search technique, coupled with the strategy of replacing a ray by a point or a line can be used to solve, in linear time, the intersection radius problem for a set ofn line segments in the plane. Currently, no algorithms are known that solve these intersection radius problems within the same time bounds.     

L 1 shortest paths among polygonal obstacles in the plane

We present an algorithm for computingL ₁ shortest paths among polygonal obstacles in the plane. Our algorithm employs the “continuous Dijkstra” technique of propagating a “wavefront” and runs in timeO(E logn) and spaceO(E), wheren is the number of vertices of the obstacles andE is the number of “events.” By using bounds on the density of certain sparse binary matrices, we show thatE =O(n logn), implying that our algorithm is nearly optimal. We conjecture thatE =O(n), which would imply our algorithm to be optimal. Previous bounds for our problem were quadratic in time and space. Our algorithm generalizes to the case of fixed orientation metrics, yielding anO(n?^?1/2 log² n) time andO(n?^?1/2) space approximation algorithm for finding Euclidean shortest paths among obstacles. The algorithm further generalizes to the case of many sources, allowing us to compute anL ₁ Voronoi diagram for source points that lie among a collection of polygonal obstacles.     

A note on a problem in the theory of grammatical complexity

In this note we answer an open question in the theory of grammatical complexity: We show that if L is an infinite context-free language, then L_n, the set of words in L of length less than or equal to n can be generated by O(|L_n|^2/3) context-free productions, and if L is an infinite linear language, then L_n can be generated by O(√|L_n|) linear productions. We also show that these bounds are the best possible.     

Optimal parallel algorithms for rectilinear link-distance problems

We provide optimal parallel solutions to several link-distance problems set in trapezoided rectilinear polygons. All our main parallel algorithms are deterministic and designed to run on the exclusive read exclusive write parallel random access machine (EREW PRAM). LetP be a trapezoided rectilinear simple polygon withn vertices. InO(logn) time usingO(n/logn) processors we can optimally compute:

1. Minimum réctilinear link paths, or shortest paths in theL ₁ metric from any point inP to all vertices ofP.
2. Minimum rectilinear link paths from any segment insideP to all vertices ofP.
3. The rectilinear window (histogram) partition ofP.
4. Both covering radii and vertex intervals for any diagonal ofP.
5. A data structure to support rectilinear link-distance queries between any two points inP (queries can be answered optimally inO(logn) time by uniprocessor).

Our solution to 5 is based on a new linear-time sequential algorithm for this problem which is also provided here. This improves on the previously best-known sequential algorithm for this problem which usedO(n logn) time and space.5 We develop techniques for solving link-distance problems in parallel which are expected to find applications in the design of other parallel computational geometry algorithms. We employ these parallel techniques, for example, to compute (on a CREW PRAM) optimally the link diameter, the link center, and the central diagonal of a rectilinear polygon.     

Exact Algorithms for the Bottleneck Steiner Tree Problem

Given n points, called terminals, in the plane ℝ² and a positive integer k, the bottleneck Steiner tree problem is to find k Steiner points from ℝ² and a spanning tree on the n+k points that minimizes its longest edge length. Edge length is measured by an underlying distance function on ℝ², usually, the Euclidean or the L ₁ metric. This problem is known to be NP-hard. In this paper, we study this problem in the L _p metric for any 1≤p≤∞, and aim to find an exact algorithm which is efficient for small fixed k. We present the first fixed-parameter tractable algorithm running in f(k)⋅nlog ² n time for the L ₁ and the L _∞ metrics, and the first exact algorithm for the L _p metric for any fixed rational p with 1<p<∞ whose time complexity is f(k)⋅(n ^k +nlog n), where f(k) is a function dependent only on k. Note that prior to this paper there was no known exact algorithm even for the L ₂ metric.     

Computing largest empty circles with location constraints

LetQ = {q₁, q₂,..., q_n} be a set ofn points on the plane. The largest empty circle (LEG) problem consists in finding the largest circleC with center in the convex hull ofQ such that no pointq _i εQ lies in the interior ofC. Shamos recently outlined anO(n logn) algorithm for solving this problem.⁽⁹⁾ In this paper it is shown that this algorithm does not always work correctly. A different approach is proposed here and shown to also result in anO(n logn) algorithm. The new approach has the advantage that it can also solve more general problems. In particular, it is shown that if the center ofC is constrained to lie in an arbitrary convexn-gon, an0(n logn) algorithm can still be obtained. Finally, an0(n logn +k logn) algorithm is given for solving this problem when the center ofC is constrained to lie in an arbitrary simplen-gonP. wherek denotes the number of intersections occurring between edges ofP and edges of the Voronoi diagram ofQ andk ?O(n ²).     

Optimal one-page tree embeddings in linear time

In the minimum linear arrangement problem one wishes to assign distinct integers to the vertices of a given graph so that the sum of the differences (in absolute value) across the edges of the graph is minimized. This problem is known to be NP-complete for the class of all graphs, but polynomial for trees—algorithms of time complexity O(n^2.2) and O(n^1.6) were given by Shiloach [SIAM J. Comput. 8 (1979) 15-32] and Chung [Comput. Math. Appl. 10 (1984) 43-60], respectively. We present a linear-time algorithm for finding the optimal embedding (arrangement) in a restricted but important class of embeddings called one-page embeddings.¹     

An optimal algorithm for the maximum-density path in a tree

We studied the problem of finding the maximum-density path in a tree. By spine decomposition and a linear-time algorithm for the maximum density segment problem, we developed an O(nlogn) time algorithm, which improves the previously best result of O(nlog²n) by using centroid decomposition. We also show the time complexity is optimal in the algebraic computation tree model.     

Minimization of the maximal lateness for a single machine

Consideration was given to the classical NP-hard problem 1|r_j|L_max of the scheduling theory. An algorithm to determine the optimal schedule of processing n jobs where the job parameters satisfy a system of linear constraints was presented. The polynomially solvable area of the problem 1|r_j|L_max was expanded. An algorithm was described to construct a Pareto-optimal set of schedules by the criteria L_max and C_max for complexity of O(n³logn) operations.     

On computing the diameter of a point set in high dimensional Euclidean space

We consider the problem of computing the diameter of a set of n points in d-dimensional Euclidean space under Euclidean distance function. We describe an algorithm that in time O(dnlogn+n²) finds with high probability an arbitrarily close approximation of the diameter. For large values of d the complexity bound of our algorithm is a substantial improvement over the complexity bounds of previously known exact algorithms. Computing and approximating the diameter are fundamental primitives in high dimensional computational geometry and find practical application, for example, in clustering operations for image databases.     

The region approach for computing relative neighbourhood graphs in the Lp metric

The following geometrical proximity concepts are discussed: relative closeness and geographic closeness. Consider a setV={v ₁,v ₂, ...,v _n } of distinct points in atwo-dimensional space. The pointv _j is said to be arelative neighbour ofv _i ifd _p (v _i ,v _j )≤max{d _p (v _j ,v _k ),d _p (v _j ,v _k )} for allv _k ∈V, whered _p denotes the distance in theL _p metric, 1≤p≤∞. After dividing the space around the pointv _i into eight sectors (regions) of equal size, a closest point tov _i in some region is called anoctant (region, orgeographic) neighbour ofv _i. For anyL _p metric, a relative neighbour ofv _i is always an octant neighbour in some region atv _i. This gives a direct method for computing all relative neighbours, i.e. for establishing therelative neighbourhood graph ofV. For every pointv _i ofV, first search for the octant neighbours ofv _i in each region, and then for each octant neighbourv _j found check whether the pointv _j is also a relative neighbour ofv _i. In theL _p metric, 1<p<∞, the total number of octant neighbours is shown to be θ(n) for any set ofn points; hence, even a straightforward implementation of the above method runs in θn ²) time. In theL ₁ andL _∞ metrics the method can be refined to a θ(n logn+m) algorithm, wherem is the number of relative neighbours in the output,n-1≤m≤n(n-1). TheL ₁ (L _∞) algorithm is optimal within a constant factor.     

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(n ²), 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/?_crit + 1)n(logn)²), wherea ≥b are the lengths of the sides of a rectangle and ?_crit 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.     

Systolic algorithms for computing the visibility polygon and triangulation of a polygonal region

This paper first presents a naive systolic algorithm for finding a closest point for each of n given points in linear time. Then, based on the algorithm, we propose linear-time systolic algorithms for the computation of the visibility polygon and for the trapezoidal partition or triangulation of a polygonal region which may contain holes. The visibility problem among n vertical line segments in the plane is also solved.     

Relative neighborhood graphs in the Li-metric

The relative neighborhood graph of a set of n points in the plane under the L₁-metric is considered. An algorithm that runs in O(nlog n) time for constructing the relative neighborhood graph based on the Delaunay triangulation is presented, improving a previously known algorithm that runs in O(n²log n) time.     

Computing turning points of curves implicitly defined by nonlinear equations depending on a parameter

Let the space curveL be defined implicitly by the (n, n+1) nonlinear systemH(u)=0. A new direct Newton-like method for computing turning points ofL is described that requires per step only the evaluation of one Jacobian and 5 function values ofH. Moreover, a linear system of dimensionn+1 with 4 different right hand sides has to be solved per step. Under suitable conditions the method is shown to converge locally withQ-order two if a certain discretization stepsize is appropriately chosen. Two numerical examples confirm the theoretical results.