B. M. E. Moret. Towards a discipline of experimental algorithmics. In M.H. Goldwasser, D.S. Johnson, and C.C. McGeoch, editors, Data Structures, Near Neighbor Searches, and Methodology: Fifth and Sixth DIMACS Implementation Challenges, DIMACS Monographs 59, pages 197--213, Providence, 2002. American Mathematical Society.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....in detail. Instead of an abstract conclusion, Section 8 collects all the rules in a check list that can possibly be used later when looking for ideas how to improve a diagram or for teaching. Related Work A number of papers on the methodology of experimental algorithmics have come out recently [10, 8, 9, 6]. In particular, 6] explains some of the rules presented here. However these papers have a wider range of topics and hence cannot be as detailed as the present paper. There are also entire books on presenting data graphically [5, 4, 17] These books explain some of the rules considered here in a ....

B. M. E. Moret. Towards a discipline of experimental algorithmics. In 5th DIMACS Challenge, DIMACS Monograph Series, 2000. to appear.

....etc. Last, but not least, encouraging fruitful cooperation between theoreticians and practitioners. 1. 2 Motivations The main motivations for recurring to experiments for analyzing and drawing onthe road conclusions about algorithms have already been pointed out by different authors [3, 16, 19, 26]. Among many, we cite the following: Many authors call Asymptopia the range of problem instances for which an algorithm exhibits clear asymptotic behavior. Unfortunately, for certain algorithms, Asymptopia may include only huge problem instances, far beyond the needs of any reasonable ....

....machines. Although it seems there is no sound and generally accepted solution to these issues, some researchers have proposed accurate and comprehensive guidelines on different aspects of the empirical evaluation of algorithms maturated from their own experience of in the field (see, for example [3, 16, 19, 26]) The interested reader may find in [24] an annotated bibliography of experimental algorithmics sources addressing methodology, tools and techniques. In the remainder of this paper, we first survey some relevant contributions to the area of Experimental Algorithmics, pointing out their impact in ....

B.M.E. Moret. Towards a discipline of experimental algorithmics. In Proceedings of the 5th DIMACS Challenge Workshop, 1996. Available over the Internet at the URL: http://www.cs.amherst.edu/ dsj/methday.html.

No context found.

Moret, B. Towards a discipline of experimental algorithmics. In Proc. 5th DIMACS Challenge Workshop (1996). Available over the Internet at the URL: http://www.cs.amherst.edu/#dsj/methday.html.

No context found.

B. Moret, "Towards a discipline of experimental algorithmics," in DIMACS Monographs in Discrete Mathematics and Theoretical Computer Science, American Mathematical Society, 2001. To appear. Available at www.cs.unm.edu/moret/dimacs.ps.

No context found.

Moret, B. Towards a discipline of experimental algorithmics. In Proc. 5th DIMACS Challenge Workshop (1996). Available over the Internet at the URL: http://www.cs.amherst.edu/#dsj/methday.html.

....leaf labelled trees on n leaves is (2n 5) and because our input space is further expanded by the choice of evolutionary rates, it is not possible to take a fair sample of the entire input space. In order to obtain statistically robust results, we followed the advice of McGeoch [18] and Moret [20] and used a number of runs, each composed of a number of trials (a trial is a single comparison) computed the mean outcome for each run, and studied the mean and standard deviation over the runs of these events. This approach is preferable to using the same total number of samples in a single ....

Moret, B.M.E., "Towards a discipline of experimental algorithmics," Proc. 5th & 6th DIMACS Implementation Challenges, M. Goldwasser, D.S. Johnson, and C.C. McGeoch, eds, American Mathematical Society, Providence, 2003. Available at www.cs.unm.edu/#moret/dimacs.ps.

....3.1 Approach Our methodology has two principal components: an approach to the conversion of PRAM algorithms into parallel programs for shared memory machines and a matching complexity model for the prediction of performance. In addition, we make use of the best precepts of algorithm engineering [34] to ensure that our implementations are as efficient as possible. Converting a PRAM algorithm to a parallel program requires us to address three problems: i) how to partition the tasks (and data) among the very limited number of processors available; ii) how to optimize locally as well as ....

.... (ii) how to optimize locally as well as globally the use of caches; and (iii) how to minimize the work spent in synchronization (barrier calls) The first two are closely related: good data and task partitioning will ensure good locality; coupling such partitioning with cache sensitive coding (see [27, 28, 29, 34] for discussions) provides programs that take best advantage of the architecture. Minimizing the work done in synchronization barriers is a fairly simple exercise in program analysis, but turns out to be far more difficult in practice: a tree based gather and broadcast barrier, for instance, will ....

B.M.E. Moret. Towards a discipline of experimental algorithmics. In DIMACS Monographs in Discrete Mathematics and Theoretical Computer Science. American Mathematical Society, 2001. To appear. Available at www.cs.unm.edu/moret/dimacs.ps.

....varying number of processors and sparse graphs with fixed size n = 8192. Fig. 2. Running times of ear decomposition on the NPACI Sun E10K with 1 to 32 pr cessor a) on var] ngpr3 lem sizes (top) and b) di#erO tsparx grr models with n = 8192 (bottom) make use of the bestrp] ts ofalgor11p engineer] [34] to ensur e that our im lementations ar e as e#cient as ossible. Conver ing a PRAM algor thm to aar[ lelrFO]F rFO ir es us to addr ess thr eer oblems: i) how to ar tition the tasks (and data) among the ver y limited number ofr ocessor s available; ii) how to o timize locally as well as globally ....

.... to o timize locally as well as globally the use of caches; and (iii) how to minimize the wor k s ent in synchr onization (bar r ier calls) The fir st two ar e closelyr elated: good data and task ar titioning willensur e good locality; cou ling such ar titioning with cache sensitive coding (see [27,28,29,34] for discussions) r videsrF15F s that take best advantage of the ar chitectur e. Minimizing the wor k done in synchr onization bar r ier s is a fair ly simpleexer cise inpr ogr am analysis, buttur ns out to be far mor e di#cult inpr actice: a tr15x ased gather]F d br oadcast barF er for instance, ....

B. M. E. Moret. Towards a discipline of experimental algorithmics. In DIMACS Monographs in Discrete Mathematics and Theoretical Computer Science. American Mathematical Society, 2001. To appear. Available at www.cs.unm.edu/#moret/dimacs.ps. 132

.... 27, 29] ever more e#cient implementation of network flow algorithms [7, 8, 13] and the characterization of the behavior of everyday algorithms and data structures such as priority queues [15, 32] shortest paths [6] minimum spanning trees [22] and sorting [21] More references can be found in [20] as well as by going to the web site for the ACM Journal of Experimental Algorithmics at www.jea.acm.org. High Performance Algorithm Engineering High Performance Algorithm Engineering focuses on one of the many facets of algorithm engineering. The high performance aspect does not immediately ....

....when designing query algorithms on these databases. While several programs in use in the area (such as BLAST, see www.ncbi.nlm.nih.gov BLAST ) have already been engineered for performance, most such e#orts have been more or less ad hoc. The emergence of a discipline of algorithm engineering [20] is bringing us a collection of tools and practices that can be applied to almost any existing algorithm or software package to speed up its execution, often by very significant factors. While we illustrated the approach and its potential results with a specific program in phylogeny reconstruction ....

Moret, B.M.E., "Towards a discipline of experimental algorithmics," Proc. 5th DIMACS Challenge (to appear), available at www.cs.unm.edu/#moret/dimacs.ps.

....the inversion distance between two signed permutations. We also provide ample experimental evidence that our linear time algorithm is efficient in practice as well as in theory: we coded it as well as the algorithm of Berman and Hannenhalli, using the best principles of algorithm engineering [14, 16] to ensure that both implementations would be as efficient as possible, and compared their running times on a large range of instances generated through simulated evolution. The two implemenations on the web are naturally far slower. The paper is organized as follows. We begin by recalling some ....

B.M.E. Moret. Towards a discipline of experimental algorithmics. In DIMACS Monographs in Discrete Mathematics and Theoretical Computer Science. American Mathematical Society, 2001. To appear. Available at www.cs.unm.edu/moret/dimacs.ps.

....represents a significant advance. 7 High Performance Computing Even the best algorithms for phylogeny reconstruction are likely to take exponential time in many cases, so that we should take advantage of high performance tools whenever possible. We used the best precepts of algorithm engineering [19] to improve the running time of our GRAPPA software, eventually achieving a 2,000fold speedup, as reported in [20] More recently, we parallelized our software (an easy task, since it offers embarrassing parallelism ) and used the 512 processor Los Lobos supercluster at the U. of New Mexico to ....

Moret, B.M.E., "Towards a discipline of experimental algorithmics," to appear in DIMACS Monographs, 2001.

No context found.

B. Moret, "Towards a discipline of experimental algorithmics," in DIMACS Monographs in Discrete Mathematics and Theoretical Computer Science, American Mathematical Society, 2001. To appear. Available at www.cs.unm.edu/#moret/dimacs.ps.

....distance between two signed permutations. We also provide ample experimental evidence that our linear time algorithm is efficient in practice as well as in theory: we coded it as well as the algorithm of Berman and Hannenhalli, using the best principles of algorithm engineering [McGeoch, 1996, Moret, 2000] to ensure that both implementations would be as efficient as possible, and compared their running times on a large range of instances generated through simulated evolution. The paper is organized as follows. We begin by recalling some definitions, briefly review past work on sorting by ....

Moret, B. (2000). Towards a discipline of experimental algorithmics. In DIMACS Monographs in Discrete Mathematics and Theoretical Computer Science, 5th DIMACS Challenge. American Mathematical Society. available at www.cs.unm.edu/moret/dimacs.ps.

No context found.

Moret, B.M.E., "Towards a discipline of experimental algorithmics," Proc. 5th DIMACS Challenge, available at www.cs.unm.edu/#moret/dimacs.ps.

....leaf labelled trees on n leaves is (2n 5) and because our input space is further expanded by the choice of evolutionary rates, it is not possible to take a fair sample of the entire input space. In order to obtain statistically robust results, we followed the advice of McGeoch [16] and Moret [18] and used a number of runs, each composed of a number of trials (a trial is a single comparison) computed the mean outcome for each run, and studied the mean and standard deviation over the runs of these events. This approach is preferable to using the same total number of samples in a single ....

Moret, B.M.E., \Towards a discipline of experimental algorithmics," to appear in Proc. 5th DIMACS Challenge, available at www.cs.unm.edu/~moret/dimacs.ps.

No context found.

Moret, B.M.E., "Towards a discipline of experimental algorithmics," Proc. 5th DIMACS Challenge, available at www.cs.unm.edu/moret/dimacs.ps.

No context found.

B. M. E. Moret. Towards a discipline of experimental algorithmics. In M.H. Goldwasser, D.S. Johnson, and C.C. McGeoch, editors, Data Structures, Near Neighbor Searches, and Methodology: Fifth and Sixth DIMACS Implementation Challenges, DIMACS Monographs 59, pages 197--213, Providence, 2002. American Mathematical Society.

No context found.

B. M. E. Moret. Towards a discipline of experimental algorithmics. In M.H. Goldwasser, D.S. Johnson, and C.C. McGeoch, editors, Data Structures, Near Neighbor Searches, and Methodology: Fifth and Sixth DIMACS Implementation Challenges, DIMACS Monographs 59, pages 197--213, Providence, 2002. American Mathematical Society.

No context found.

B. Moret. Towards a discipline of experimental algorithmics. In 5th DIMACS Challenge, DIMACS Monograph Series, 2000.

No context found.

B. M. E. Moret, \Towards a discipline of experimental algorithmics,", to appear in Proc. 5th DIMACS Challenge, C.C. McGeoch and D.S. Johnson, eds. DIMACS Monograph Series, AMS Press, 2000.

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