P. Flajolet, B. Salvy, and P. Zimmermann. Automatic Average-Case Analysis of Algorithms. Theor. Comp. Sci., (79):37--109, 1991.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....the first two sections and the generating function consequences are considered in the third section. The fourth section introduces the Maple implementation and is followed by a comparison with the LUO system for automatic average case algorithm analysis developed by the Algorithms Project of INRIA [5, 10]. 1.1. Grammars and Specifications. The structures considered here are those described in detail in [5] They are built from atoms of weight one (commonly expressed as Z) and o, ffls of weight 0. The constructors are disjoint union j, cartesian product Delta, sequence , Set( MultiSet( and ....

....section introduces the Maple implementation and is followed by a comparison with the LUO system for automatic average case algorithm analysis developed by the Algorithms Project of INRIA [5, 10] 1.1. Grammars and Specifications. The structures considered here are those described in detail in [5]. They are built from atoms of weight one (commonly expressed as Z) and o, ffls of weight 0. The constructors are disjoint union j, cartesian product Delta, sequence , Set( MultiSet( and Cycle( These latter constructors form, respectively, sequences, sets without or with duplication and ....

[Article contains additional citation context not shown here]

P. Flajolet, B. Salvy, and P. Zimmermann. Automatic average-case analysis of algorithms. Theoretical Computer Science, Series A, 79(1):37--109, February 1991.

....size. Accommodating this requires manual or semi automatic transformation of the time or space function [17, 29] The analysis is mainly asymptotic. A challenging problem that arises in this approach is optimizing the time bound or space bound function to a closed form in terms of the input size [17, 24, 7]. But closed forms are known only for subclasses of functions. Thus, such optimization can not be done automatically for analyzing general programs. Rosendahl proposed characterizing inputs using partially known input structures [24] For example, instead of replacing an input list l with its ....

P. Flajolet, B. Salvy, and P. Zimmermann. Automatic average-case analysis of algorithms. Theoretical Computer Science, Series A, 79(1):37-109, Feb. 1991.

....object grammar. It is most often described with drawings. For instance, the standard decomposition of complete binary trees is an object grammar (Figure 1) The formalism given here for object grammars [7, 8] generalizes the one for context free grammars. It is akin to the work of Flajolet al... [10, 11] allowing for the specication of structures by grammars involving set, sequence and cycle constructions. One can also categorize object grammars as belonging to the domain of Universal Algebra and Magmas [9, 15, 1] Finally, our approach is related to the Theory of Species [3, 14] which gives a ....

P. Flajolet, B. Salvy, and P. Zimmermann. Automatic average-case analysis of algorithms. Theoretical Computer Science, 79(1):37109, 1991.

.... it was first developed by Gross [24] Subsequent developments are due to: Cori and Richard [9] leading to the enumeration of planar maps [7] Fliess [9] with applications to solving differential equations; Viennot [36] with applications in enumerative combinatorics; Flajolet [22], with applications to asymptotic analysis. This method is now known as the DSV methodology, following M.P. Schtzenberger s wish expressed to Viennot [35] In 1975, a theoretical computer science workshop was held in France, on the topic of formal power series. Its proceedings, edited by J. ....

....= 1 pE) x Ez ) 1 mx mx xE and we obtain the generating function for the number e n of well formed arithmetic expressions having n symbols by setting all the variables equal to x: E(x) e n x = n0 1 6x x 4 4x . This example is borrowed from the L W cookbook [22]. Recall that L U W is the software developed by the Flajolet s team at INRIA. It computes automatically asymptotic expressions for the coefficients of a series. More classical is the computation on the Dyck language which leads to the equation D = x xD 1. An elementary calculation shows ....

P. Flajolet, B. Salvy, P. Zimmerman, Automatic average-case analysis of algorithms, Theor. Comp. Sci. 79, 1 (1991), 37-109.

....e.g. 4] but it would be very di#cult to manually design appropriate models at the required fine grain level. Indeed in program analysis, the timing semantics can hardly be designed at the programming language source level (for which automatic concrete com# plexity analysis is certainly useful [47] but insu#cient since constants factors do matter [7] In practice it is indispensable to consider the program semantics at the assembler level, that is for a given compiler and for a given processor with some hypotheses on the frequency of physical interrupts [81] The model being automatically ....

P. Flajolet, B. Salvy, and P. Zimmermann. Automatic average-case analysis of algorithm. Theoret. Comput. Sci. , 79(1):37--109, 1991.

....Performance 1. INTRODUCTION Analysis of program running time is important for reactive systems, interactive environments, compiler optimizations, performance evaluation, and many other computer applications. It has been extensively studied in many fields of computer science: algorithms [22, 13, 14, 41], programming languages [39, 23, 32, 36, 35] and systems [37, 30, 34, 33] Being able to predict accurate time bounds automatically and efficiently is particularly important for many applications, such as reactive systems. It is also particularly desirable to be able to do so for high level ....

....independent of the input size, i.e. are constants while the computation iterates or recurses. Whatever values of the primitive parameters are assumed, a second problem arises, and it is theoretically challenging: optimizing the time bound function to a closed form in terms of the input size [39, 6, 23, 32, 14]. Although much progress has been made in this area, closed forms are known only for subclasses of functions. Thus, such optimization can not be automatically done for analyzing general programs. In systems, inputs are characterized indirectly using loop bounds or execution paths in programs, and ....

[Article contains additional citation context not shown here]

P. Flajolet, B. Salvy, and P. Zimmermann. Automatic average-case analysis of algorithms. Theoretical Computer Science, Series A, 79(1):37--109, Feb. 1991.

....and formal language theory has a long story. Probabilistic tools have been used early in information theory [11] automata theory [5] and coding theory [3] 8] More recently, evidence of the usefulness of this interaction have been found in fields like asymptotic analysis of algorithms [6], allocation problems, concurrency measures [7] communication protocols specification and verification, and random generation of combinatorial structures. Most of these approaches rely on describing various subsets of A , the set of words on a finite alphabet A, defining probability measures on ....

Flajolet, P., Salvy, B., Zimmermann, P., Automatic Average-Case Analysis of Algorithms, Theoretical Computer Science 79, 1991, 37-109.

....this requires manual or semiautomatic transformation of the time or space function [18, 29] The analysis is mainly asymptotic. A theoretically challenging problem that arises in this approach is optimizing the time bound or space bound function to a closed form in terms of the input size [18, 24, 6]. Although much progress has been made in this area, closed forms are known only for subclasses of functions. Thus, such optimization can not be automatically done for analyzing general programs. 1 Rosendahl proposed characterizing inputs using partially known input structures [24] For ....

P. Flajolet, B. Salvy, and P. Zimmermann. Automatic average-case analysis of algorithms. Theoretical Computer Science, Series A, 79(1):37-109, Feb. 1991.

....this requires manual or semi automatic transformation of the time or space function [53, 32, 55] The analysis is mainly asymptotic. A theoretically challenging problem that arises in this approach is optimizing the time bound or space bound function to a closed form in terms of the input size [53, 5, 32, 45, 12]. Although much progress has been made in this area, closed forms are known only for subclasses of functions. Thus, such optimization can not be automatically done for analyzing general programs. Rosendahl proposed characterizing inputs using partially known input structures [45] For example, ....

....is easy to automatically generate all binary trees of height h with unknown elements or all binary trees with n nodes and unknown elements. 9 Related work There has been a large amount of work on analyzing program cost or resource complexities, but the majority of it is on time analysis, e.g. [53, 32, 11, 45, 51, 49, 12, 44, 33, 47, 34]. Some techniques for time analysis can 28 of 31 be adapted for space analysis, for example, as we did for stack space and heap allocation analysis. Analysis of live heap space has an important di erence from all these other analyses: it involves explicit analysis of the graph structure of the ....

P. Flajolet, B. Salvy, and P. Zimmermann. Automatic average-case analysis of algorithms. Theoretical Computer Science, Series A, 79(1):37-109, Feb. 1991.

....G omez s address: Computer Science Department, Indiana University, Bloomington, IN 47405 7104. Corresponding author: Yanhong A. Liu. Email: liu cs.sunysb.edu. Tel: 631 632 8463. Fax: 631 632 8334. URL: http: www.cs.sunysb.edu liu . 1 studied in many elds of computer science: algorithms [25, 16, 17, 53], programming languages [50, 26, 41, 44] and systems [46, 37, 43, 42] It is particularly important for many applications, such as real time systems and embedded systems, to be able to predict accurate time bounds and space bounds automatically and eciently, and it is particularly desirable to be ....

....independent of input size, i.e. are constants while the computation iterates or recurses. Whatever values of the primitive cost parameters are assumed, a second problem arises, and it is theoretically 3 challenging: optimizing the cost bound function to a closed form in terms of the input size [50, 10, 26, 41, 17, 7]. Although much progress has been made in this area, closed forms are known only for subclasses of functions. Thus, such optimization can not be automatically done for analyzing general programs. In systems, inputs are characterized indirectly using loop bounds or execution paths in programs, and ....

[Article contains additional citation context not shown here]

P. Flajolet, B. Salvy, and P. Zimmermann. Automatic average-case analysis of algorithms. Theoretical Computer Science, Series A, 79(1):37-109, Feb. 1991.

....corresponding generating functions. They are present, implicitly or explicitly, in most books that cover combinatorial enumeration, such as [81, 173, 351, 377] The most systematic approach to developing and using general rules of this type has been carried out by Flajolet and his collaborators [139]. They develop ways to see immediately (cf. 134] that if we consider mappings of a set of n labeled elements to itself, so that all n n distinct mappings are considered equally likely, then the generating function for the longest path length is given by f(z) # # k=0 # 1 1 t(z) e ....

....is precisely what computer systems do well. It is therefore possible to write programs that will derive the asymptotics behavior from the specification of the recurrence. More generally, one can analyze asymptotics of a much greater variety of generating functions. Flajolet, Salvy, and Zimmermann [124, 139] have written a powerful program for just such computations. Their system uses Maple to carry out most of the basic analytic computations. It contains a remarkable amount of automated expertise in recognizing generating functions, computing their singularities, and extracting asymptotic ....

[Article contains additional citation context not shown here]

P. Flajolet, B. Salvy, and P. Zimmermann, Automatic average--case analysis of algorithms, Theoretical Computer Science, 79 (1991), 37--109.

....function consequences are considered in the third section, including a treatment of pathlength. The fourth section introduces the Maple implementation and is followed by a comparison with the LUO system for automatic average case algorithm analysis developed by the Algorithms Project of INRIA [5, 12]. There is a brief consideration of random generation in section seven. 1 2 MARNI MISHNA 1.1. Grammars and Specifications. The structures considered here are built from basic objects of weight 1, an Atom (commonly expressed as Z) and of weight 0, an ffl. The constructors are disjoint union j, ....

....of steps (however that is defined) an algorithm requires when a given structure is input. Generating functions summarize this information and offer a means for automatic average case complexity analysis. This idea has been explored in depth with respect to the decomposable structures defined here [4, 5, 12]. In fact, the system LUO is an implementation of this concept for a family of algorithms. The main drawback of this system is that only the final univariate generating functions are available. Attribute grammars give relations for multivariate generating functions and hence for problems it can ....

P. Flajolet, B. Salvy, and P. Zimmermann. Automatic average-case analysis of algorithms. Theoretical Computer Science, Series A, 79(1):37--109, February 1991.

....have a number of useful and general tools available in the form of Mellin transforms and iteration theory of analytic functions. 6. Automatic Analysis The approach of finding general decidable asymptotic properties of combinatorial structures has been prolonged. Flajolet, Salvy and Zimmermann [1] have designed a system called Lambda Upsilon Omega ( Upsilon Omega ) that implements a number of decision procedures on combinatorial structures like the ones discussed here. The kernel specification language consists of the constructions of union, product, sequence, sets, multisets and cycles ....

Flajolet (P.), Salvy (B.), and Zimmermann (P.). -- Automatic average-case analysis of algorithms. Theoretical Computer Science, Series A, vol. 79, n 1, February 1991, pp. 37--109.

....to estimate the number of solutions of database query [25] Their method estimates the query size dynamically at run time, in contrast our method is a static analysis performed at compile time. Much of the work on automatic complexity analysis is in the context of functional programming languages [12, 15, 21, 34, 37, 41, 44]. We extend their work by being able to handle nondeterminism and 34 the generation of multiple solutions via backtracking in logic programs. Kaplan considers the analysis of the average case complexity of logic programs [18] but his approach cannot handle programs that can produce multiple ....

P. Flajolet, B. Salvy and P. Zimmermann, "Automatic average-case analysis of algorithms," Theoretical Computer Science, 79 (1991), pp. 37--109.

....just an approximation, but also a bound on the actual execution cost. Fortunately, as mentioned before, much work has been presented on (time) complexity analysis of programs (Le M etayer 1988, Wadler 1988, Rosendhal 1989, Bjerner and Holmstrom 1989, Sarkar 1989, Zimmermann and Zimmermann 1989, Flajolet et al. 1991). The most directly applicable are the methods presented by Debray and Lin (1993) and Debray et al. 1994) for statically estimating cost functions for predicates in a logic program. The two approaches have much in common but they differ in the way the approximation is done. In the first one upper ....

Flajolet, P., Salvy, B., Zimmermann, P. (1991). Automatic Average-Case Analysis of Algorithms. Theor. Comp. Sci., (79):37--109.

....lazy approach) Combinatorics and the analysis of algorithms. In combinatorics, the analysis of algorithms and for the random generation of combinatorial objects, one usually needs to expand generating functions up to a high order. Therefore, this is the ideal application for relaxed power series (Flajolet et al. 1990). In this context, multivariate power series correspond to the study of parameters in enumeration problems or the analysis of a certain algorithm (Soria, 1990) Numerical analysis. In numerical analysis, power series are mainly computed in order to be evaluated. The required number of terms ....

Flajolet, P., Salvy, B., Zimmermann, P. (1990). Automatic average-case analysis of algorithms. T.C.S., 79(1):37-109.

....this requires manual or semi automatic transformation of the time or space function [53, 32, 55] The analysis is mainly asymptotic. A theoretically challenging problem that arises in this approach is optimizing the time bound or space bound function to a closed form in terms of the input size [53, 5, 32, 45, 12]. Although much progress has been made in this area, closed forms are known only for subclasses of functions. Thus, such optimization can not be automatically done for analyzing general programs. Rosendahl proposed characterizing inputs using partially known input structures [45] For example, ....

....time by only a constant factor. The running times of all three bound functions grow at the same rate as the corresponding space functions. 8 Related work There has been a large amount of work on analyzing program cost or resource complexities, but the majority of it is on time analysis, e.g. [53, 32, 11, 45, 51, 49, 12, 44, 33, 47, 34]. Some techniques for time analysis can be adapted for space analysis, for example, as we did for stack space and heap allocation analysis. Analysis of live heap space has an important difference from all these other analyses: it involves explicit analysis of the graph structure of the data. Most ....

P. Flajolet, B. Salvy, and P. Zimmermann. Automatic average-case analysis of algorithms. Theoretical Computer Science, Series A, 79(1):37--109, Feb. 1991.

....contain, at least in principle, all the necessary information needed to predict an algorithm s behaviour. This part is strongly influenced by an approach introduced earlier for the automatic analysis of some classes of algorithms over decomposable structures by Flajolet, Salvy, and Zimmermann [11]. The real dimension of generating functions lies in their complex analytic properties, especially when we contemplate them near singularities. Without this aspect, the collection of formal generating function equations would remain somewhat devoid of content. A systematic analysis of ....

....lies in their complex analytic properties, especially when we contemplate them near singularities. Without this aspect, the collection of formal generating function equations would remain somewhat devoid of content. A systematic analysis of singularities, at either a finite or infinite distance [10, 11], permits us to extract in all cases of practical interest the asymptotic costs involved. For instance, two different strategies, nicknamed little endian and big endian , when applied to Cayley trees (non plane labelled trees) lead to an average complexity of the form O(n ) and It is to ....

[Article contains additional citation context not shown here]

Flajolet, P., Salvy, B., and Zimmermann, P. Automatic average--case analysis of algorithms. Theoretical Computer Science, Series A 79, 1 (Feb. 1991), 37--109.

.... robustness (as defined above) and the density of the graph 2 (i.e. the number of its edges) The originality of our approach consists in introducing in this range of problems methods of analytic combinatorics [14, 21] and recent research in automatic analysis based on symbolic computation [8, 13, 15, 23]. Additional threshold estimates regarding properties of multiple sourcedestination pairs are discussed in the last section of the paper. Summary of results. From earlier known results [7, 20] and this paper, a picture of robustness under the G n;p model emerges. As is usual in random graph ....

....into smaller structures either of the same type or of simpler types and then in extracting, from such a decomposition, the corresponding recurrence relations. The approach developed here is direct and symbolic , as it relies on a precise specification language for combinatorial structures [13, 15]. It is based on so called admissible constructions that have the important feature of admitting direct translations into generating functions. Let A be a class of combinatorial objects with an associated notion of size. We let A n denote 1 the subset of objects in A that have size n and write ....

[Article contains additional citation context not shown here]

Ph. Flajolet, B. Salvy, and P. Zimmermann, "Automatic Average--case Analysis of Algorithms", Theoretical Computer Science 79(1), pp. 37--109, 1991.

....with a focus on order constraints and analysis of algorithms. Flajolet and Steyaert developed rules initially specialized to trees from which a complexity calculus could be derived for a wide class of algorithms [22, 24, 36, 37, 75] This was later extended into a much more general system [32] to be discussed in Part III. We propose now to explain the major principle of a symbolic approach to the derivation of generating functions. 5 1. Trains. 2. Formal specification. 8 : train = locomotive wagons) wagons = ....

....is compiled from the specification [x3] The generating function is then solved explicitly [x4] Currently, the analysis of this problem can be achieved automatically. A system, Lambda Upsilon Omega ( Upsilon Omega ) has been designed by B. Salvy and P. Zimmermann jointly with the author [32, 69, 84]. It does the analysis and via an implementation of complex asymptotic methods and singularity analysis, it is also able to find automatically the asymptotic form of the coefficients: The number of trains of size n satisfies the estimate trainn n 0:07097007911 Delta 1:930298068 n : 6 ....

[Article contains additional citation context not shown here]

Flajolet, P., Salvy, B., and Zimmermann, P. Automatic average--case analysis of algorithms. Theoretical Computer Science, Series A 79, 1 (February 1991), 37--109. 22

No context found.

P. Flajolet, B. Salvy, and P. Zimmermann. Automatic Average-Case Analysis of Algorithms. Theor. Comp. Sci., (79):37--109, 1991.

No context found.

P. Flajolet, B. Salvy, and P. Zimmermann. Automatic Average-Case Analysis of Algorithms. Theor. Comp. Sci., (79):37-109, 1991.

No context found.

Flajolet, P., Salvy, B., and Zimmermann, P. Automatic average-case analysis of algorithms. Theoretical Computer Science 79, 1 (1991), 37--109.

No context found.

Flajolet, P., Salvy, B., Zimmermann, P.: Automatic Average-Case Analysis of Algorithms. Theoretical Computer Science 79 (1991) 37-109

No context found.

Flajolet, P., Salvy, B., and Zimmermann, P. Automatic average-case analysis of algorithms. Theoretical Computer Science 79, 1 (1991), 37--109.

First 50 documents

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