J. JaJa. An Introduction to Parallel Algorithms. Addison-Wesley, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....communication is a primitive of the model. 4 Providing Shared Memory on D BSP A very desirable feature of a distributed memory model is the ability to support a shared memory abstraction efficiently. Among the other benefits, this feature allows porting the vast body of PRAM algorithms [J aJ92] to the model at the cost of a small time penalty. In this section we present a number of results that demonstrate that D BSP can be endowed with an efficient shared memory abstraction. Implementing shared memory calls for the development of a scheme to represent m shared cells (variables) among ....

J. J aJ a. An Introduction to Parallel Algorithms. Addison Wesley, Reading MA, 1992.

....O(n) messages in sequential succession. Second, as the search progresses, only one processor at a time is busy working, while the others sit idle. Thus, there would be no parallel speedup. In fact, no ecient parallel implementations are known at this point for depth rst graph traversal methods [7]. 3.2 A Possible Approach Our novel approach attempts to minimize the communication while improving the load balancing by keeping as many processors as possible busy with useful work. There are three major phases in the algorithm. During the rst phase, each processor examines its local subgraph ....

J. JaJa. An Introduction to Parallel Algorithms. Addison-Wesley Publishing Company, New York, 1992.

.... [9] B log M cache faults general [14] skew P (L gn P ) log 3 n log P log(n=P ) time general [12] P L log time general skew time integer skew 4 time, O(n log n) work general [12] O log time, O(n log n) work general skew arbitrary CRCW PRAM [25] O(log n) time, O(n) work (rand. constant [13] time, O(n) work (rand. constant skew Similarly, a; b) a; b 1] and s[a; b) s[a; b 1] The operator is used for the concatenation of strings. Consider a string s = s[0; n) over the alphabet = 1; n] The sux array SA contains the suxes ....

J. Jaja. An Introduction to Parallel Algorithms. Addison Wesley, 1992.

....however, such parallelizations are not always efficient. Leslie Grignard s Fast Multipole Algorithm. Floating Point Systems FFT (Fast Fourier Transform an algorithm for computing discrete Fourier transforms) Tree based reductions, particularly for the PRAM model, as described in [J 92] Certain well known algorithms for solving the N body problem, for example the Barnes Hut algorithm and some algorithms of John Salmon. It is interesting to note that just because an algorithm is based on a (sequential) divideand conquer strategy does not mean that it must be parallelized ....

J. J aJ a. An Introduction to Parallel Algorithms. Addison-Wesley, 1992.

....arithmetic operations. With a divide and conquer implementation of Euclid s algorithm we can achieve the second step using only O(2nlog (2n) operations. An implementation on parallel computers lets us perform the two steps in only O(log(2n) time and O(log (2n) time respectively (see [13]) The study of the stability is based upon the condition number of the Hankel submatrices. This subject is still under development (see [21] However, there are forward methods that enable us to jump over ill conditioned submatrices and most of them can be implement via the backward scheme. 4. ....

J. J' aj' a, An Introduction to Parallel Algorithms, Addisson-Wesley, Reading, Mass., 1992.

....T [i 2 ; j 1 ; j 1 km 1] 6= T [i 2 1; j 1 ; j 1 3 km 1] and T [i 1 ; j 1 ; j 1 km 1] 6= T [i 1 1; j 1 ; j 1 km 1] When k = 1, we say simply block. 2. 4 The lowest common ancestor problem (LCA) De nition 3 (Lowest common ancestor) The lowest common ancestor [14,15,16] of two vertices u and v of a rooted tree is the vertex w that is an ancestor of u and v and that is farthest from the root. We denote w = LCA(u; v) De nition 4 (Lowest common ancestor problem) The lowest common ancestor problem consists of pre processing a rooted tree such that queries of the ....

....such that queries of the type LCA(u; v) for any vertices u and v of the tree, can be answered in constant sequential time. 2.5 Range Minima Problem De nition 5 Given a vector of n real numbers A = a 1 ; a 2 ; an ) de ne MIN(i; j) minfa i ; a j g. The range minima problem [14,8,9] consists of pre processing vector A such that queries MIN(i; j) for any 1 i j n, can be answered in constant time. 2.6 Cartesian Trees and k height Cartesian Trees De nition 6 (Cartesian tree) Let A 0;n 1 = a 0 ; a 1 ; an 1 ) be an array of n distinct real numbers, the ....

J. JaJa, An Introduction to Parallel Algorithms (Addison-Wesley, Reading, 1992).

....graph problems [6] In this work we consider the range minima problem. We present an algorithm under the CGM(n; p) model that solves the problem requiring a constant number of communication rounds and O(n=p) computation time. In the PRAM model, there is an optimal algorithm of O(log n) time [13]. This algorithm, however, is not immediately transformable to the CGM model. Berkman et al. 4] describe an optimal algorithm of O(log log n) time for a PRAM CRCW. This algorithm motivates the CGM algorithm presented in this paper. We do not, however, compare the performance of the proposed ....

....Berkman et al. 4] experimentally. This is because Berkman s PRAM algorithm uses a large number of processors (n= log log n) as opposed to the proposed algorithm, which is more realistic in terms of the number of processors. Range minima is a basic problem and appears in many graph problems. In [13], range minima is used together with Euler tour in the design of a PRAM algorithm for the problem of lowest common ancestor (LCA) LCA, in turn, is used in such graph algorithms as open ear decomposition, biconnectivity, and planarity problems [17] In addition to the LCA problem, range minima is ....

[Article contains additional citation context not shown here]

J. Jaja, An Introduction to Parallel Algorithms (Addison-Wesley Publishing Company, 1992).

....a consequence, the preprocessing stage involves computation of simpler functions and needs simpler data structures. Interval maximum can be found in constant time by applying a standard technique that uses precomputed tables of total size O(n log n) The tables store pre x minima and sux maxima [7]. We explain how to arrange these tables in such a way that F (u; v) can be found using two table lookups for nding the JP order, one xor operation, one operation nding the most signi cant nonzero bit, two table lookups in fused pre x and sux tables and some shifts and adds for index ....

J. Jaja. An Introduction to Parallel Algorithms. Addison Wesley, 1992.

....of phases is O( n) using a simple criterion, and O(n ) for a more refined criterion with high probability (whp) Sect. 4 presents an adaption of the phase driven approach to the CRCW PRAM model which allows p processors (PUs) concurrent read write access to a shared memory in unit cost (e.g. [13]) We propose an algorithm for random graphs with random edge weights that runs in O(n log n) time whp. The work, i.e. the product of its running time and the number of processors, is bounded by O(n log n dn) whp. In Sect. 5 we adapt the basic idea to external memory computation (I O model ....

J. J'aj'a. An Introduction to Parallel Algorithms. Addison-Wesley, 1992.

....etc. Di#erent application areas call for di#erent correctness properties. In a controller of an embedded system, e.g. error outcomes like stack overflow or outof memory are totally unacceptable. For such applications a compiler essentially has to preserve total correctness [MO97,MOW99] This is di#erent for strictly transformational programs. The crucial requirement for this program class is that a regular results of an implementation is always a possible results of the source program because this allows to rely on results without further scrutiny. This property essentially ....

....requirement for this program class is that a regular results of an implementation is always a possible results of the source program because this allows to rely on results without further scrutiny. This property essentially amounts to preservation of partial correctness (PPC) MO95a,MO96b,GMO96,MOW99] and is vital in bootstraps of verified compilers, as compilers are transformational programs. In [MO96b] we characterise PPC (in the situation where the state sets of source and target program may di#er) in various semantic styles and prove that these characterisations are equivalent. Immediate ....

[Article contains additional citation context not shown here]

Markus Muller-Olm and Andreas Wolf. On excusable and inexcusable failures: Towards an adequate notion of translation correctness. In FM'99, Lecture Notes in Computer Science. Springer, 1999. to appear.

No context found.

J. JaJa. An Introduction to Parallel Algorithms. Addison-Wesley, 1992.

No context found.

J. J aJ a (1992) An introduction to parallel algorithms. Addison-Wesley, Reading.

No context found.

J. JaJa. Introduction to parallel algorithms. Addison-Wesley, Reading, Mass., 1992.

No context found.

J. J aJ a. An Introduction to Parallel Algorithms. Addison-Wesley, 1992.

No context found.

J aj a, J. An Introduction to Parallel Algorithms. Addison Wesley, 1992.

No context found.

J aj a, J.: An Introduction to parallel algorithms. Addison Wesley, 1992.

No context found.

J. JaJa. An Introduction to Parallel Algorithms. Addison-Wesley, 1992.

No context found.

J. JaJa, An Introduction to Parallel Algorithms. Addison-Wesley, 1992.

No context found.

J aj a, J. An Introduction to Parallel Algorithms. Addison Wesley, 1992.

No context found.

J aj a, J.: An Introduction to parallel algorithms. Addison Wesley, 1992.

No context found.

J. Jaja, An Introduction to Parallel Algorithms, Addison-Wesley, 1992.

No context found.

J. JaJa, An Introduction to Parallel Algorithms, Addison-Wesley, 1992.

No context found.

J. Jaja. An Introduction to Parallel Algorithms. Addison Wesley, 1992.

No context found.

J. JaJa, An Introduction to Parallel Algorithms, Addison-Wesley, 1992.

No context found.

J. JaJa, An Introduction to Parallel Algorithms, Addison-Wesley, Reading, MA, 1992.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC