M. Hermenegildo, G. Puebla, K. Marriott, and P. Stuckey. Incremental Analysis of Logic Programs. In Proc. of the Int. Conf. on Logic Programming, Tokyo, Japan, MIT Press, June 1995, 797--814.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....andagents (and agents are processing agents working in and parallel with each other) Dependencies are detected at run time by executing some simple tests in troduced by the parallelizing compiler. ACE adopts the technique originally designed by DeGroot (DeGroot 1984) and refined by Hermenegildo (Hermenegildo et al. 1995) of annotating the program at compile time with Conditional Graph Expressions (CGEs) h conditions i ) B 1 Delta Delta Delta Bn ) where h conditions i is a conjunction of simple tests on variables appearing in the clause that verifies whether the arguments share any variables with ....

....is extremely low (around 5 ) The automatic annotation performed by the ACE compiler was rather slow, due to the size and organization of the application a single module of over 35; 000 lines of Prolog code. A modular reorganization of the code and the use of incremental analysis techniques (Hermenegildo et al. 1995) will improve the speed of annotation. The annotator is written in Prolog, so the annotation process itself can be parallelized. Comparison with Other Work Experience shows that NLP applications are highly parallel in nature. Although considerable research has been proposed in using parallelism ....

Hermenegildo, M. et al. 1995. Incremental analysis of logic programs. In ICLP95. MIT Press.

....based on such semantics would require considerably more machinery than that given here. The problem of incremental analysis of logic programs, where analysis can be carried out even if the program being analyzed is not available in its entirety, has been investigated by Hermenegildo et al. [24]. While the underlying motivation for this work resembles ours in many ways, the details differ substantially. In particular, the approach of Hermenegildo et al. involves re analyzing (parts of) a program in response to changes to the program, while our approach involves first computing the ....

M. Hermenegildo, G. Puebla, K.Marriott, and P.J. Stuckey. Incremental Analysis of Logic Programs. In Proc. Twelfth International Conference on Logic Programming, pages 797--811.

....for ULTRA System The automatic annotation performed by the ACE compiler was rather slow, due to the size and organization of the application a single module of over 35; 000 lines of Prolog code. A modular reorganization of the code and the use of a newly developed incremental analysis technique (Hermenegildo et al. 1995) will improve the speed of annotation. Even more, the annotator is written in Prolog, so the annotation process itself can be parallelized. Comparison with Other Work Experience shows that NLP applications are highly parallel in nature. Although considerable research has been proposed in using ....

Hermenegildo, M. et al. 1995. Incremental analysis of logic programs. In ICLP95. MIT Press.

.... of analysis) is achieved by considering an abstract domain D with certain characteristics (such as being nite, or of nite height, or without in nite ascending chains) or by the use of a widening operator [15] We do not describe here how to build AO(P; p; D ) Details can be found in [3, 45, 48, 25]. The graph has two sorts of nodes: those which correspond to literals (called or nodes) and those which correspond to clauses (called and nodes) Or nodes are triples hp i ; c i ; s i i and the set of or nodes in AO(P; p; D ) Analysis(P; p; D ) And nodes are also triples hH j ....

....Generate Code Optimize Code Final Parallelization Program Abstract Domain FIGURE 6.1. Program Parallelization and Abstract Multiple Specialization system uses incremental analysis techniques to eciently obtain updated analysis information from the one generated for the original program [25, 52]. Conceptually, the process of abstract multiple specialization is composed of ve steps, which are shown in Figure 6.1 (picture on the right) In the rst step (simplify) the program optimizations based on abstract execution are performed whenever possible. This saves having to optimize the ....

[Article contains additional citation context not shown here]

M. Hermenegildo, G. Puebla, K. Marriott, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming, pages 797{

No context found.

M. Hermenegildo, G. Puebla, K. Marriott, and P. Stuckey. Incremental Analysis of Logic Programs. In Proc. of the Int. Conf. on Logic Programming, Tokyo, Japan, MIT Press, June 1995, 797--814.

No context found.

M. Hermenegildo, K. Marriott, G. Puebla and P. Stuckey. Incremental analysis of logic programs. Proc. Twelfth Int. Conf. on Logic Programming, pages 797--811, MIT Press, 1995.

No context found.

M. Hermenegildo, G. Peubla, K. Marriott, and P. Stuckey. Incremental Analysis of Logic Programs. In Proc. of the Int. Conf. on Logic Programming, Tokyo, Japan, MIT Press, June 1995, 797--814.

No context found.

M. Hermenegildo, G. Puebla, K. Marriott, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming, pages 797-811. MIT Press, June 1995.

No context found.

M. Hermenegildo, G. Puebla, K. Marriott, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming, pages 797--811. MIT Press, June 1995.

No context found.

M. Hermenegildo, K. Marriott, G. Puebla and P. Stuckey. Incremental Analysis of Logic Programs. To appear in Proc. of 1995.

No context found.

M. Hermenegildo, G. Puebla, K. Marriott, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming, pages 797-811. MIT Press, June 1995.

....is in the way they organize to redo all computations dependent on an invalid table lookup : the way they attempt to minimize the number of transitions to be redone and how they attempt to make the best use of what has already been computed. Our PLAI implementation of the fixpoint algorithm [Hermenegildo et al. 1995; Muthukumar and Hermenegildo 1989; 1990; 1992] is performed as follows. The program is preprocessed in order to determine recursive predicates and recursive rules. This allows analyzing nonrecursive predicates in one pass without checking whether there is an ancestor node. For the recursive ....

Hermenegildo, M., Marriott, K., Puebla, G., and Stuckey, P. 1995. Incremental analysis of logic programs. In Proceedings of the 12th International Conference on Logic Programming, L. Sterling, Ed. MIT Press, Cambridge, Mass., 797--811.

....compiler for CIAO is principally concerned with automatic parallelization of programs, and performs optimization of built in predicates, so the optimizations performed are completely different from those described herein. The optimizer also supports multivariant specialization (see Puebla and Hermenegildo [1995,1997] and makes use of incremental analysis [Hermenegildo et al. 1995] The results from the CIAO system support our findings of the usefulness of mul ACM Transactions on Programming Languages and Systems, Vol. 8, No. 1, January 1999. 114 Delta A. Kelly et al. Prin = principal, Time = no. ....

....of programs, and performs optimization of built in predicates, so the optimizations performed are completely different from those described herein. The optimizer also supports multivariant specialization (see Puebla and Hermenegildo [1995,1997] and makes use of incremental analysis [Hermenegildo et al. 1995]. The results from the CIAO system support our findings of the usefulness of mul ACM Transactions on Programming Languages and Systems, Vol. 8, No. 1, January 1999. 114 Delta A. Kelly et al. Prin = principal, Time = no. of months, Rate = annual interest rate, MP = monthly payment, B ....

[Article contains additional citation context not shown here]

Hermenegildo, M., Puebla, G., Marriott, K., and Stuckey, P. 1995. Incremental analysis of logic programs. In Logic Programming: Proceedings of the 12th International Conference, L. Sterling, Ed. MIT Press, Cambridge, Mass., 797--814.

....the program usually gets small modifications and is checked again for inconsistency with the requirements. In such a case a non incremental analyser repeats much of the work performed previously. We will thus aim at obtaining analysis tools which are incremental, using the algorithms presented in [HPMS95] 5.6 Comparators Properties of the program inferred by static analysis may be very informative and may be used for debugging purposes. The programmer may read the analysis information and compare with what he expected. However, this may be a tedious task as analysis information may be hard to ....

M. Hermenegildo, G. Puebla, K. Marriott, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming, pages 797--811. MIT Press, June 1995.

.... Such global analysis has also been applied to concurrent constraint programming systems, where one of the most important objectives is to reduce suspension and resumption of goals and synchronization overhead [9, 33, 34, 30, 35, 17] Recent progress in incremental and compositional global analysis [33, 23, 4] appears to solve most remaining problems related to dealing with large programs and the interactive program development environment that is common in constraint programming systems. However, the application of extensive optimization in commercial or widely used public domain systems still remains ....

M. Hermenegildo, G. Puebla, K. Marriott, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming, pages 797-811. MIT Press, June 1995.

....de ned in the source language, is supported [HPB99] 6 tively, the abstract call and success substitutions. 7 Due to space limitations, and given that it is now well understood, we do not describe here how we compute Analysis(P; entries; D ) More details can be found in [Bru91, MH90, MH92, HPMS95, PH96] Given Analysis(P; entries; D ) fhp 1 ; c 1 ; s 1 i; hp n ; c n ; s n ig, correctness of abstract interpretation guarantees that the following propositions hold: Proposition 3.2 [Correctness w.r.t. successes] The abstract success substitutions cover all the ....

....correct until a distributed xed point is reached. The ineciency of possibility 1 is less dramatic when incremental analysis is used. In that case, the previous analysis results for the module are used in order to compute the new local xed point. For this, the incremental addition algorithm of [HPMS95] can be used. This is not a great restriction since incremental analysis algorithms can be as fast as non incremental ones [PH96] Possibility 2 has the advantage of not performing any speculative work as analysis does not continue until an optimal success substitution is computed. However, care ....

M. Hermenegildo, G. Puebla, K. Marriott, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming, pages 797-811. MIT Press, June 1995.

....PLAI implements a generic (goaldependent and goal independent) top down driven abstract interpreter. The whole computation is domain independent. This allows plugging in different abstract domains, provided suitable interfacing functions are defined. PLAI also incorporates incremental analysis [HPMS95] in order to deal with large programs and is capable of analyzing full languages (in particular, full standard Prolog [BCHP96, CRV94] A modification of the PLAI framework capable of analyzing dynamically scheduled programs is also provided in order to support the concurrent models. Note that, ....

M. Hermenegildo, G. Puebla, K. Marriott, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming, pages 797--811. MIT Press, June 1995.

.... for efficient fixpoint computation, a subject that has received considerable attention [CC77, MH90, MH92, LDMH93, VWL94, Jor94] Incremental analysis of logic programs has been shown to be feasible and much more efficient in certain contexts than traditional (non incremental) global analysis [HMPS95, KB95]. In particular, HMPS95] discussed the different types of changes that had to be dealt with in an incremental setting, provided overall solutions for dealing with such changes (in terms of which parts of the analysis graph need to be updated and recomputed) and proposed a basic set of solutions ....

.... a subject that has received considerable attention [CC77, MH90, MH92, LDMH93, VWL94, Jor94] Incremental analysis of logic programs has been shown to be feasible and much more efficient in certain contexts than traditional (non incremental) global analysis [HMPS95, KB95] In particular, [HMPS95] discussed the different types of changes that had to be dealt with in an incremental setting, provided overall solutions for dealing with such changes (in terms of which parts of the analysis graph need to be updated and recomputed) and proposed a basic set of solutions that showed the ....

[Article contains additional citation context not shown here]

M. Hermenegildo, K. Marriott, G. Puebla, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming. MIT Press, June 1995.

....debugging and validation and Section 4 for program transformation and optimization. 2 Static Analysis and Program Assertions The fundamental functionality behind Ciaopp is static program analysis. For this task Ciaopp uses the PLAI abstract interpreter [18, 2] its CLP [11] and incremental [15, 22] versions, and adaptations of Gallagher s regular type analysis [8] The system includes several abstract domains and can infer information on basic properties such as moded types, de niteness, freeness, and grounding dependencies, as well as on more complex properties such as determinacy, ....

M. Hermenegildo, G. Puebla, K. Marriott, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming, pages 797-811. MIT Press, June 1995.

....still correct, meaning that in practice little additional analysis is needed. Analysis of a hypothetical goal just uses information already available in the analyzer except where new calling patterns are created, which causes some additional work. Details about the algorithms used can be found in [2]. Details of the description domains are deliberately kept insulated from the optimizer so as to make it easier to change them. Descriptions of five of the six analysis domains can be found in [6] The new domain, Bounds, gives an interval for each arithmetic variable in which the variable s ....

M. Hermenegildo, G. Puebla, K. Marriott, and P. Stuckey. Incremental analysis of logic programs. In Proc. of the Int. Conf. on Logic Programming, Tokyo, Japan, MIT Press, June 1995, 797--814.

....can be used as approximations in order to compute a more precise success substitution (see [2] for more details) Dynamic analysis and optimization. There is still another, quite different and interesting solution to the problem of dynamic predicates, which is based on incremental global analysis [17]. Note that in order to implement assert some systems include a copy of the full compiler at run time. The idea would be to also include the (incremental) global analyzer and the analysis information for the program, computed for the static part of the program. The program is in principle ....

....(and reoptimized) This has the advantage of having fully optimized code at all times, at the cost of increasing the cost of calls to database manipulation predicates and of executable size. A system along these lines has been built by us for a parallelizing compiler. The results presented in [17] show that such a reanalysis can be made in a very small fraction of the normal compilation time. 6 Program Modules The main problem with studying the impact of modularity in analysis (and the reason we have left the issue until this section) is the lack of even a de facto standard. There have ....

[Article contains additional citation context not shown here]

M. Hermenegildo, K. Marriott, G. Puebla, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming. MIT Press, 1995.

....modification of the analyzer, as described previously. The analysis time overhead resulting from this modification has been measured at 3 on the average, which we argue is quite tolerable. Furthermore, the same modification is used for other purposes, most notably for incremental global analysis [16], and is therefore now in any case a permanent addition to the analyzer. Only one pass of the analyzer is required to generate both the multiply specialized program and to obtain the information needed to determine the optimizations applicable to each version. These optimizations are of the ....

....the total time needed would be the sum of both times. However, as mentioned above, during the automatic parallelization process, an analysis is generally done to optimize the run time tests. This first analysis can in fact be reused for the multiple specialization with a few modifications [16]. For each benchmark program we present the analysis time, the multiple specialization time, their sum, and the percentage of the total time used in specialization. All the times are in milliseconds and have been measured on a SPARC 10. We argue that the time required for multiple specialization, ....

M. Hermenegildo, K. Marriott, G. Puebla, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming. MIT Press, June 1995.

....PLAI implements a generic (goal dependent and goal independent) top down driven abstract interpreter. The whole computation is domain independent. This allows plugging in different abstract domains, provided suitable interfacing functions are defined. PLAI also incorporates incremental analysis [HMPS95] in order to deal with large programs and is capable of analyzing full languages (in particular, full standard Prolog [BCHP96, CRH94] A modification of the PLAI framework capable of analyzing dynamically scheduled programs is also provided in order to support the concurrent models. Note that, ....

M. Hermenegildo, K. Marriott, G. Puebla, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming. MIT Press, June 1995.

....is performed by means of abstract interpretation [14, 17] The underlying framework of analysis is that of PLAI [42, 54, 56] PLAI implements a generic, top down driven abstract interpreter. It allows easily plugging into it several abstract domains. PLAI also incorporates incremental analysis [38] in order to deal with large programs, and is capable of analyzing full languages (in particular, full standard Prolog [2, 11] CIAO also includes a version of the PLAI framework which is capable of accurately analyzing (constraint) programs with dynamic scheduling [49, 31] This also allows ....

M. Hermenegildo, K. Marriott, G. Puebla, and P. Stuckey. Incremental Analysis of Logic Programs. In International Conference on Logic Programming. MIT Press, June 1995.

....by the dependency arcs from the initial calls S and marks those entries in the dependency arc and answer table which are reachable. The remainder are removed. Space limitations prevent us from providing examples for this algorithm or for the incremental versions proposed they can be found in [10]. It is also important to remember the purpose of this algorithm. It is not intended as a practical algorithm for computing a program analysis graph, as the overhead of event handling is too incremental addition(R) foreach rule Ak Bk;1 ; Bk;n k 2 R foreach entry A : CP 7 AP in the ....

....of additional code required to handle incremental addition. The results are shown in Table 1. All execution times are milliseconds on a Sparc 10. A relatively wide range of programs has been used as benchmarks. Due to lack of space, they are not discussed here (they are described in more detail in [10, 3] and can be obtained from http: www.clip.dia.fi.upm.es) However, the number of clauses is included in the table (column Cl) for reference. Strd is the time taken by PLAI s original fixpoint in order to analyze the whole program as one block. Incr is the equivalent figure for the slightly ....

M. Hermenegildo, K. Marriott, G. Puebla, and P. Stuckey. Incremental Analysis of Logic Programs. Technical Report CLIP 14/94.0, Computer Science Dept., Technical U. of Madrid (UPM), Spain, October 1994. ftp://clip.dia.fi.upm.es/pub/papers/

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