Efficient geometric algorithms on the EREW PRAM

We present a technique that can be used to obtain efficient parallel geometric algorithms in the EREW PRAM computational model. This technique enables us to solve optimally a number of geometric problems in O(log n) time using O(n/log n) EREW PRAM processors, where n is the input size of a problem. These problems include: computing the convex hull of a set of points in the plane that are given sorted, computing the convex hull of a simple polygon, computing the common intersection of half-planes whose slopes are given sorted, finding the kernel of a simple polygon, triangulating a set of points in the plane that are given sorted, triangulating monotone polygons and star-shaped polygons, and computing the all dominating neighbors of a sequence of values. PRAM algorithms for these problems were previously known to be optimal (i.e., in O(log n) time and using O(n/log n) processors) only on the CREW PRAM, which is a stronger model than the EREW PRAM     

An EREW PRAM algorithm for image component labeling

An important midlevel task for computer vision is addressed. The problem consists of labeling connected components in N^1/2 ×N^2/2 binary images. This task can be solved with parallel computers by using a simple and novel algorithm. The parallel computing model used is a synchronous fine-grained shared-memory model where only one processor can read from or write to the same memory location at a given time. This model is known as the exclusive-read exclusive-write parallel RAM (EREW PRAM). Using this model, the algorithm presented has O(log N) complexity. The algorithm can run on parallel machines other than the EREW PRAM. In particular, it offers an optimal image component labeling algorithm for mesh-connected computers     

Parallel Heap Operations on an EREW PRAM

We consider parallel heap operations on the exclusive-read exclusive-write parallel random-access machine. We first present an O(n/p + log p) time parallel algorithm to construct a heap of n elements using p processors, which is optimal for p θ(n/log n). We then propose a parallel root deletion algorithm. In a preparatory step, a data structure for dynamic processor allocation is constructed using O((n/log n)^{1 − 1/k}) processors in O(log k) time for some constant k, 1 ≤ k ≤ ⌈log(n/log n)⌉. A sequence of root deletions can then be performed, each of which takes O((log n)/p + log p + log log n) time using p processors. Finally, we discuss a parallel algorithm running in O((log n)/p + log p) time for inserting an element into a heap, which is optimal for p = θ((log n)/log log n). Both deletion and insertion algorithms run in O(log log n) time when p = θ((log n)/log log n).     

Maintaining B-Trees on an EREW PRAM

Efficient and practical algorithms for maintaining general B-trees on an EREW PRAM are presented. Given a B-tree of order b with m distinct records, the search (respectively, insert and delete) problem for n input keys is solved on an n-processor EREW PRAM in O(log n + b log_b m) (respectively, O(b(log n + log_b m)) and O(b²(log_b n + log_b m))) time.     

Fast integer merging on the EREW PRAM

We investigate the complexity of merging sequences of small integers on the EREW PRAM. Our most surprising result is that two sorted sequences ofn bits each can be merged inO(log logn) time. More generally, we describe an algorithm to merge two sorted sequences ofn integers drawn from the set {0, ...,m?1} inO(log logn+log min{n, m}) time with an optimal time-processor product. No sublogarithmic-time merging algorithm for this model of computation was previously known. On the other hand, we show a lower bound of Ω(log min{n, m}) on the time needed to merge two sorted sequences of lengthn each with elements drawn from the set {0, ...,m?1}, implying that our merging algorithm is as fast as possible form=(logn)^Ω(1). If we impose an additional stability condition requiring the elements of each input sequence to appear in the same order in the output sequence, the time complexity of the problem becomes Θ(logn), even form=2. Stable merging is thus harder than nonstable merging.     

A randomized sublinear time parallel GCD algorithm for the EREW PRAM

We present a randomized parallel algorithm that computes the greatest common divisor of two integers of n bits in length with probability 1−o(1) that takes O(nloglogn/logn) time using O(n6+?) processors for any ?>0 on the EREW PRAM parallel model of computation. The algorithm either gives a correct answer or reports failure.We believe this to be the first randomized sublinear time algorithm on the EREW PRAM for this problem.     

Improved Bounds for Integer Sorting in the EREW PRAM Model

A new simple method of exploiting nonstandard word length in the nonconservative RAM and PRAM models is considered. As a result, improved bounds for parallel integer sorting in the EREW PRAM model with standard and nonstandard word length are obtained.     

A Randomized Linear-Work EREW PRAM Algorithm to Find a Minimum Spanning Forest

We present a randomized EREW PRAM algorithm to find a minimum spanning forest in a weighted undirected graph. On an n -vertex graph the algorithm runs in o(( log n) ^{1+ ɛ} ) expected time for any ɛ >0 and performs linear expected work. This is the first linear-work, polylog-time algorithm on the EREW PRAM for this problem. This also gives parallel algorithms that perform expected linear work on two general-purpose models of parallel computation—the QSM and the BSP.     

PARALLEL GIVENS SEQUENCES FOR SOLVING THE GENERAL LINEAR MODEL ON A EREW PRAM∗

Abstract

Parallel Givens sequences for solving the General Linear Model (GLM) are developed and analyzed. The block updating GLM estimation problem is also considered. The solution of the GLM employs as a main computational device the Generalized QR Decomposition, where one of the two matrices is initially upper triangular. The proposed Givens sequences efficiently exploit the initial triangular structure of the matrix and special properties of the solution method. The complexity analysis of the sequences is based on a Exclusive Read-Exclusive Write (EREW) Parallel Random Access Machine (PRAM) model with limited parallelism. Furthermore, the number of operations performed by a Givens rotation is determined by the size of the vectors used in the rotation. With these assumptions one conclusion drawn is that a sequence which applies the smallest number of compound disjoint Givens rotations to solve the GLM estimation problem does not necessarily have the lowest computational complexity. The various Givens sequences and their computational complexity analyses will be useful when addressing the solution of other similar factorization problems.     

A Randomized Linear-Work EREW PRAM Algorithm to Find a Minimum Spanning Forest

We present a randomized EREW PRAM algorithm to find a minimum spanning forest in a weighted undirected graph. On an n -vertex graph the algorithm runs in o(( log n)1+?) expected time for any ? >0 and performs linear expected work. This is the first linear-work, polylog-time algorithm on the EREW PRAM for this problem. This also gives parallel algorithms that perform expected linear work on two general-purpose models of parallel computation—the QSM and the BSP.     

Work-time optimal k-merge algorithms on the PRAM

For 2⩽k⩽n, the k-merge problem is to merge a collection of ksorted sequences of total length n into a new sorted sequence. The k-merge problem is fundamental as it provides a common generalization of both merging and sorting. The main contribution of this work is to give simple and intuitive work-time optimal algorithms for the k-merge problem on three PRAM models, thus settling the status of the k-merge problem. We first prove that Ω(n log k) work is required to solve the k-merge problem on the PRAM models. We then show that the EREW-PRAM and both the CREW-PRAM and the CRCW require Ω(log n) time and Ω(log log n+log k) time, respectively, provided that the amount of work is bounded by O(n log k). Our first k-merge algorithm runs in Θ(log n) time and performs Θ(n log k) work on the EREW-PRAM. Finally, we design a work-time optimal CREW-PRAM k-merge algorithm that runs in Θ(log log n+log k) time and performs Θ(n log k) work. This latter algorithm is also work-time optimal on the CREW-PRAM model. Our algorithms completely settle the status of the k-merge problem on the three main PRAM models     

Efficient PRAM simulation on a distributed memory machine

We present algorithms for the randomized simulation of a shared memory machine (PRAM) on a Distributed Memory Machine (DMM). In a PRAM, memory conflicts occur only through concurrent access to the same cell, whereas the memory of a DMM is divided into modules, one for each processor, and concurrent accesses to the same module create a conflict. Thedelay of a simulation is the time needed to simulate a parallel memory access of the PRAM. Any general simulation of anm processor PRAM on ann processor DMM will necessarily have delay at leastm/n. A randomized simulation is calledtime-processor optimal if the delay isO(m/n) with high probability. Using a novel simulation scheme based on hashing we obtain a time-processor optimal simulation with delayO(log log(n) log*(n)). The best previous simulations use a simpler scheme based on hashing and have much larger delay: (log(n)/log log(n)) for the simulation of an n processor PRAM on ann processor DMM, and (log(n)) in the case where the simulation is time-processor optimal.Our simulations use several (two or three) hash functions to distribute the shared memory among the memory modules of the PRAM. The stochastic processes modeling the behavior of our algorithms and their analyses based on powerful classes of universal hash functions may be of independent interest.Research partially supported by NSF/DARPA Grant CCR-9005448. Work was done while at the University of California at Berkeley and the International Computer Science Institute, Berkeley, CA.Research partially supported by National Science Foundation Operating Grant CCR-9016468, National Science Foundation Operating Grant CCR-9304722, United States-Israel Binational Science Foundation Grant No. 89-00312, United States-Israel Binational Science Foundation Grant No. 92-00226, and ESPRIT BR Grant EC-US 030.Part of work was done during a visit at the International Computer Science Institute at Berkeley; supported in part by DFG-Forschergruppe Effiziente Nutzung massiv paralleler Systeme, Teilprojekt 4, and by the Esprit Basic Research Action Nr. 7141 (ALCOM II).     

Efficient and Scalable PRAM Algorithms for Discrete-Event Simulation of Bounded Degree Networks

We describe two new parallel algorithms, one conservative and another optimistic, for discrete-event simulation on an exclusive-read exclusive-write parallel random-access machine (EREW PRAM). The target physical systems are bounded degree networks which are represented by logic circuits. Employing p processors, our conservative algorithm can simulate up to O(p) independent messages of a system with n logical processes in O(log n) time. The number of processors, p, can be optimally varied in the range 1 ≤ p ≤ n. To identify independent messages, this algorithm also introduces a novel scheme based on a variable size time window. Our optimistic algorithm is designed to reduce the rollback frequency and the memory requirement to save past states and messages. The optimistic algorithm also simulates O(p) earliest messages on a p-processor computer in O(log n) time. To our knowledge, such a theoretical efficiency in parallel simulation algorithms, conservative or optimistic, has been achieved for the first time.     

Efficient algorithms for clique problems

The k-clique problem is a cornerstone of NP-completeness and parametrized complexity. When k is a fixed constant, the asymptotically fastest known algorithm for finding a k-clique in an n-node graph runs in O(n0.792k) time (given by Nešet?il and Poljak). However, this algorithm is infamously inapplicable, as it relies on Coppersmith and Winograd's fast matrix multiplication.We present good combinatorial algorithms for solving k-clique problems. These algorithms do not require large constants in their runtime, they can be readily implemented in any reasonable random access model, and are very space-efficient compared to their algebraic counterparts. Our results are the following:

•

We give an algorithm for k-clique that runs in O(nk/(εlogn)^k−1) time and O(nε) space, for all ε>0, on graphs with n nodes. This is the first algorithm to take o(nk) time and O(nc) space for c independent of k.

•

Let k be even. Define a k-semiclique to be a k-node graph G that can be divided into two disjoint subgraphs U={u₁,…,uk/2} and V={v₁,…,vk/2} such that U and V are cliques, and for all i?j, the graph G contains the edge {ui,vj}. We give an time algorithm for determining if a graph has a k-semiclique. This yields an approximation algorithm for k-clique, in the following sense: if a given graph contains a k-clique, then our algorithm returns a subgraph with at least 3/4 of the edges in a k-clique.

     

Efficient algorithms for common transversals

Suppose we are given a set S of n (possibly intersecting) simple objects in the plane such that, for every pair of objects in S, the intersection of the boundaries of these two objects has O(1) connected components. We consider the problem of determining whether there exists a straight line that goes through every object in S. We give an O(n log n γ (n)) time algorithm for this problem, where γ(n) is a very slowly growing function of n. In many cases, our algorithm runs in O(n log n) time. Previously, only special cases of this problem were considered: the case when every object is a straight-line segment (Edelsbrunner et al., 1982), the case when the objects are equal-radius circles (Bajaj and Li, 1983), and the case when objects all maintain the same orientation (Edelsbrunner, 1985). All these cases follow from our general approach, which places no constraints on the size and/or configuration of the objects in S.     

Efficient algorithms for array redistribution

Dynamic redistribution of arrays is required very often in programs on distributed presents efficient algorithms for redistribution between different cyclic(k) distributions, as defined in High Performance Fortran. We first propose special optimized algorithms for a cyclic(x) to cyclic(y) redistribution when x is a multiple of y, or y is a multiple of x. We then propose two algorithms, called the GCD method and the LCM method, for the general cyclic(x) to cyclic(y) redistribution when there is no particular relation between x and y. We have implemented these algorithms on the Intel Touchstone Delta, and find that they perform well for different array sizes and number of processors     

Efficient parallel algorithms for graph problems

We present an efficient technique for parallel manipulation of data structures that avoids memory access conflicts. That is, this technique works on the Exclusive Read/Exclusive Write (EREW) model of computation, which is the weakest shared memory, MIMD machine model. It is used in a new parallel radix sort algorithm that is optimal for keys whose values are over a small range. Using the radix sort and known results for parallel prefix on linked lists, we develop parallel algorithms that efficiently solve various computations on trees and unicycular graphs. Finally, we develop parallel algorithms for connected components, spanning trees, minimum spanning trees, and other graph problems. All of the graph algorithms achieve linear speedup for all but the sparsest graphs.Part of this work was done while the first author was at the University of Illinois, Urbana-Champaign, the second author was at Carnegie-Mellon University, and the third author was at the Hebrew University and the Courant Institute of Mathematical Sciences, New York University. A preliminary version of this work was presented at the 1986 International Conference on Parallel Processing.     

Efficient parallel algorithms for graph problems

We present an efficient technique for parallel manipulation of data structures that avoids memory access conflicts. That is, this technique works on the Exclusive Read/Exclusive Write (EREW) model of computation, which is the weakest shared memory, MIMD machine model. It is used in a new parallel radix sort algorithm that is optimal for keys whose values are over a small range. Using the radix sort and known results for parallel prefix on linked lists, we develop parallel algorithms that efficiently solve various computations on trees and “unicycular graphs.” Finally, we develop parallel algorithms for connected components, spanning trees, minimum spanning trees, and other graph problems. All of the graph algorithms achieve linear speedup for all but the sparsest graphs.     

Efficient algorithms for large-scale temporal aggregation

The ability to model time-varying natures is essential to many database applications such as data warehousing and mining. However, the temporal aspects provide many unique characteristics and challenges for query processing and optimization. Among the challenges is computing temporal aggregates, which is complicated by having to compute temporal grouping. We introduce a variety of temporal aggregation algorithms that overcome major drawbacks of previous work. First, for small-scale aggregations, both the worst-case and average-case processing time have been improved significantly. Second, for large-scale aggregations, the proposed algorithms can deal with a database that is substantially larger than the size of available memory. Third, the parallel algorithm designed on a shared-nothing architecture achieves scalable performance by delivering nearly linear scale-up and speed-up, even at the presence of data skew. The contributions made in this paper are particularly important because the rate of increase in database size and response time requirements has out-paced advancements in processor and mass storage technology.