C. C. McGeoch. Experimental analysis of algorithms. In P. Pardalos and E. Romeijn, editors, Handbook of Global Optimization, Volume 2: Heuristic Approaches. Kluwer Academic Publishers, 2002.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... in terms of randomly generated instances of SAT problems, e.g. 28, 11] or structured instances, such as the instances from the DIMACS set [30] or the SATPLAN set [23] Merits of either approach are subject to on going critique and examination [10] 24] 26] in particular, and [21] 22] [25] in general. The papers [21] and [22] succinctly articulate the case for careful experimental design an approach adopted for the experiments with SAT problems in this paper. The experimental design methodology presumes availability of well defined classes of experimental subjects. In our ....

C. C. McGeoch, Experimental Analysis of Algorithms, in Handbook of Global Optimization, Volume 2: Heuristic Approaches, P. Pardalos and E. Romeijn, ed., Kluwer Academic Publishers, 2001.

.... Papadimitriou, 1992) for the greedy algorithm for satisfiability do not apply to interesting and hard region of problems as described in x3. In addition, actual behaviour on real problems is sometimes quite different to worst and average case analyses. We therefore support the calls of McGeoch (McGeoch, 1986), Hooker (Hooker, 1993) and others for the development of an empirical science of algorithms. In such a science, experiments as well as theory are used to advance our understanding of the properties of algorithms. One of the aims of this paper is to demonstrate the benefits of such an empirical ....

....predictions. 1. Informal observations to this effect were made by Bart Selman during the presentation of (Selman et al. 1992) at AAAI 92. These observations were enlarged upon in (Gent Walsh, 1992) 49 Gent Walsh In our experiments, we followed three methodological principles from (McGeoch, 1986). First, we performed experiments with large problem sizes and many repetitions, to reduce variance and allow for emergent properties. Second, we sought good views of the data. That is, we looked for features of performance which are meaningful and which are as predictable as possible. Third, we ....

McGeoch, C. (1986). Experimental Analysis of Algorithms. Ph.D. thesis, Carnegie Mellon University. Also available as CMU-CS-87-124.

....results presented in [7] for the greedy algorithm for satisfiability do not apply to interesting and hard region of problems as described in x3. In addition, actual behaviour on real problems is sometimes quite different to worst and average case analyses. We therefore support the calls of McGeoch [9], Hooker [6] and others for the development of an empirical science of algorithms. In such a science, experiments as well as theory are used to advance our understanding of the properties of algorithms. One of the aims of this paper is to demonstrate the benefits of such an empirical approach. We ....

....these features may not be the most immediately obvious to record. Third, data must be analysed, not simply measured. Suitable analysis of data may show features which are not clear from a simple (graphical) presentation. Invaluable discussion of all these research principles is contained in [9]. In the rest of this paper we show how these principles enable us to make very detailed numerical predictions about GSAT s search. Many features of GSAT s search space can be graphically illustrated by plotting how they vary during a try. The most obvious feature to plot is the score, the number ....

C.C. McGeoch. Experimental Analysis of Algorithms. PhD thesis, Carnegie Mellon University, 1986. Also available as CMU-CS-87-124.

....for NP complete problems such as satisfiability and constraint satisfaction. Where we have not followed these maxims, we have suffered as a result. 1 Introduction The empirical study of algorithms is a relatively immature field with many technical and scientific problems. We support the calls of McGeoch (1986,1996) Hooker (1994) and others for a more scientific approach to the empirical study of algorithms. Our contribution in this paper is colloquial. We admit to a large number of mistakes in conducting our research. While painful, we hope that this will encourage others to avoid these mistakes, ....

C. McGeoch. 1986. Experimental Analysis of Algorithms. Ph.D. Dissertation, Carnegie Mellon University. Also available as CMUCS -87-124.

....a relatively immature field with many technical and scientific problems. We shall not address the many technical problems (e.g. determining appropriate performance measures and representative problem samples) but will instead look at some of the scientific problems. We support the calls of McGeoch [10], Hooker [9] and others for a more scientific approach to the empirical study of algorithms. Our contribution here is colloquial. With time and more such tales from the front line, we hope that standard experimental practices in the study of algorithms will develop which represent good science . ....

C. C. McGeoch. Experimental Analysis of Algorithms. PhD thesis, Carnegie Mellon University, 1986. Also available as CMU-CS-87-124.

....its ability to vary implementation types. In a modest amount of code using combinations of heuristics and methods, 21 variations of the push relabel algorithm are included. These variations are parameterized so that the user can experiment effectively. For a review of experimental analysis see [18, 17]. 4 LINK Applications A LINK application can be built by accessing the necessary LINK libraries and then adding application specific user interface code and libraries. A prototype application, LINKGUI, was developed in tandem with LINK. Currently, several other applications are under development ....

C. McGeoch. Experimental Analysis of Algorithms. PhD thesis, Carnegie Mellon University, August 1986.

....while Chapters 5 and 6 apply the techniques in the context of parallel algorithms design. The key idea that this dissertation tries to emphasize is the use of a systematic methodology to study parallel algorithms from both a theoretical and experimental viewpoint. Both Bentley [Ben91] and McGeoch [McG86] have presented similar methodologies for the experimental study of sequential algorithms. This thesis will present case studies that combine careful experiments with new or known theoretical results into a coherent understanding of the performance of proximity algorithms on sequential, vector, ....

....achieve good performance on a real machine. These results will be presented in a standard benchmark style. In addition, implementations, or simulations of them, can be used to gain a more detailed understanding of the algorithms themselves. McGeoch s thesis on experimental algorithms analysis [McG86] and recent work by Bentley on the traveling salesman problem [Ben89] presents a framework for experimental analysis of algorithms. In this thesis, we will apply these methods to the design and analysis of parallel algorithms. Chapter 2 presents these ideas in more detail. 1.3 Contributions ....

[Article contains additional citation context not shown here]

C. McGeoch. Experimental Analysis of Algorithms. PhD thesis, School of Computer Science, CMU, 1986.

....integer modulus a power of 2 is very cheap. In addition, we do not need to perform the subtraction longhand since the bits of (a xor b xor a Gamma b) indicate whether a borrow was necessary into a given bit position when computing a Gamma b. 6 Experiments As in previous experimental studies [ McGeoch, 1986 ] we pack objects with pseudo random integer weights into bins that are twice their maximum size. We generate n objects each with a weight drawn uniformly and randomly from (0; l] and pack into bins of capacity 2l. Exploratory tests with this model show large variation in problem difficulty. ....

....for optimization procedures [ Korf, 1995 ] The pruning rules Korf used are the analogues of the simple pruning rule in Cdbf without modular pruning. McGeoch has performed extensive experiments on the First Fit, Best Fit, Decreasing First Fit and Decreasing Best Fit approximation algorithms [ McGeoch, 1986 ] using objects with integer weights and bins of capacity 2 30 Gamma 1. She observed a critical region in which Decreasing First Fit gave packings with a large amount of empty space. As these packings tended to occur when there was a statistical excess of objects with large weights, this ....

C. C. McGeoch. Experimental Analysis of Algorithms. PhD thesis, Carnegie Mellon University, 1986. Also available as CMU-CS-87-124.

....to minimize the possibility of later misinterpretation, one should not rely on the syntax of the output file to tell you which data item is which, but should include the name of the item ( algorithm version, etc. along with each item. For more detailed discussions of these issues, see [Ben91, McG01] Principle 5. Use Reasonably Efficient Implementations This is surprisingly a somewhat controversial principle. Although it would at first glance seem obvious that we should want to use efficient implementations (especially in a field like 13 algorithm design where efficiency is one of our ....

C. C. McGeoch. Experimental analysis of algorithms. In P. Pardalos and E. Romeijn, editors, Handbook of Global Optimization, Volume 2: Heurstic Approaches. Kluwer Academic, 2001. To appear.

No context found.

C. C. McGeoch. Experimental analysis of algorithms. In P. Pardalos and E. Romeijn, editors, Handbook of Global Optimization, Volume 2: Heuristic Approaches. Kluwer Academic Publishers, 2002.

No context found.

McGeoch, C. C. 1986a. Experimental analysis of algorithms, Ph.D. thesis, CMU-CS-87124, Computer Science Department, Carnegie Mellon University, Pittsburgh, PA 15213 USA.

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