E. Albert, S. Antoy, and G. Vidal. Measuring the Eectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of the 10th Int'l Workshop on Logic-Based Program Synthesis and Transformation (LOPSTR 2000.  Home/Search   Document Not in Database   Summary   Related Articles   Check

No context found.

E. Albert, S. Antoy, and G. Vidal. Measuring the Eectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of the 10th Int'l Workshop on Logic-Based Program Synthesis and Transformation (LOPSTR 2000.

No context found.

E. Albert, S. Antoy, and G. Vidal. Measuring the E#ectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of LOPSTR 2000, pages 103--124. Springer LNCS 2042, 2001.

No context found.

E. Albert, S. Antoy, and G. Vidal. Measuring the Eectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of LOPSTR 2000, pages 103-124. Springer LNCS 2042, 2001.

No context found.

E. Albert, S. Antoy, and G. Vidal. Measuring the Eectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of the 10th Int'l Workshop on Logic-Based Program Synthesis and Transformation (LOPSTR 2000.

.... of Curry (in contrast to [10,12] which contain only partial de nitions) One can use it to prove the correctness of implementations by further re nements, as done in [21] Furthermore, one can count the costs (time space) associated to particular computations in order to justify optimizations [1,4,22] or to compare di erent search strategies. The paper is organized as follows. In the next section we introduce some foundations for understanding the subsequent development. Section 3 recalls a semantic description for lazy functional logic programs in natural style. This is re ned in Section 4 ....

E. Albert, S. Antoy, and G. Vidal. Measuring the Eectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of the 10th Int'l Workshop on Logic-based Program Synthesis and Transformation (LOPSTR'00), pages 103-124. Springer LNCS 2042.

.... Indeed, we are working on the implementation of an interpreter for Curry based on such an extended operational description covering all the aforementioned features [2] The complete operational description could be used, e.g. as a basis to define a cost augmented semantics in the style of [1,3,19,21], to develop debugging and optimization tools (like partial evaluators) and to check or derive new implementations (like in [20] for Curry. ....

E. Albert, S. Antoy, and G. Vidal. Measuring the E#ectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of the 10th Int'l Workshop on Logic-based Program Synthesis and Transformation (LOPSTR'00), pages 103--124. Springer LNCS 2042.

....the symbolic cost criteria is a good approximation of the real speedup measured experimentally. A prototype implementation of a profiler for the language Curry [19] has been also developed. B. Cost Augmented Partial Evaluation An alternative approach to the use of profiling tools is presented in [1], 36] They introduce the scheme of a narrowing driven partial evaluator enhanced with the computation of symbolic costs. While the previous approach of [4] presents two independent processes narrowing driven partial evaluation and the computation of profiling results , 1] 36] fully ....

....is presented in [1] 36] They introduce the scheme of a narrowing driven partial evaluator enhanced with the computation of symbolic costs. While the previous approach of [4] presents two independent processes narrowing driven partial evaluation and the computation of profiling results , [1], 36] fully integrate the computation of quantitative information into the specialization technique. Thus, we have available a setting in which one can discuss the e#ects of the program transformer in a precise framework and, moreover, to quantify these e#ects. This scheme may serve as a basis to ....

[Article contains additional citation context not shown here]

E. Albert, S. Antoy, and G. Vidal. Measuring the E#ectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of the 10th Int'l Workshop on LogicBased Program Synthesis and Transformation (LOPSTR 2000.

.... list and x:xs a list with rst element x and tail xs) append eval flex append [ ys = ys The evaluation annotation eval flex declares append as a exible function which can also be used to solve equations over functional expressions (see below) For instance, the equation append p s = [1,2,3] is solved by instantiating the variables p and s to lists so that their concatenation results in the list [1,2,3] The basic operational semantics of our source language is based on a combination of needed narrowing and residuation [16] The residuation principle is based on the idea of ....

.... eval flex declares append as a exible function which can also be used to solve equations over functional expressions (see below) For instance, the equation append p s = 1,2,3] is solved by instantiating the variables p and s to lists so that their concatenation results in the list [1,2,3]. The basic operational semantics of our source language is based on a combination of needed narrowing and residuation [16] The residuation principle is based on the idea of delaying function calls until they are ready for a deterministic evaluation (by rewriting) Residuation preserves the ....

[Article contains additional citation context not shown here]

E. Albert, S. Antoy, and G. Vidal. Measuring the Eectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of LOPSTR 2000, pages 103-124. Springer LNCS 2042, 2001.

.... of Curry (in contrast to [11, 13] which contain only partial definitions) One can use it to prove the correctness of implementations by further refinements, as done in [22] Furthermore, one can count the costs (time space) associated to particular computations in order to justify optimizations [1, 3, 23] or to compare di#erent search strategies. 2 Foundations A main motivation for this work is to provide a foundation for developing programming tools (like profilers, debuggers, optimizers) for declarative multiparadigm languages. In order to be concrete, we consider Curry [11, 13] as our source ....

E. Albert, S. Antoy, and G. Vidal. Measuring the E#ectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of LOPSTR 2000, pages 103--124. Springer LNCS 2042, 2001.

.... of Curry (in contrast to [10, 12] which contain only partial definitions) One can use it to prove the correctness of implementations by further refinements, as done in [19] Furthermore, one can count the costs (time space) associated to particular computations in order to justify optimizations [1, 5, 20] or to compare di#erent search strategies. The paper is organized as follows. In the next section we introduce some foundations for understanding the subsequent development. Section 3 recalls a semantic description for functional logic programs in natural style. This is refined in Section 4 to a ....

E. Albert, S. Antoy, and G. Vidal. Measuring the E#ectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of the 10th Int'l Workshop on Logic-based Program Synthesis and Transformation (LOPSTR'00), pages 103--124. Springer LNCS 2042, 2001.

....a formal basis to reason about implementationoriented aspects of programs, e.g. to develop appropriate tracing, profiling, and debugging tools. For instance, one can instrument this semantics in order to count the costs (time space) associated to particular computations (similarly to, e.g. [1,4,21,24]) This is useful to formally quantify the improvements achieved by a concrete program optimization and to compare di#erent search strategies. Note that this approach would not be possible by considering a non deterministic semantics, since it cannot properly describe the computation paths ....

....features with an example. Consider the following rule defining a function to concatenate two lists (where [ denotes the empty list and z:zs a list with first element z and tail zs) conc(xs, ys) case xs of ys; z : zs) z : conc(zs, ys) Now, the equational constraint conc(p,s) [1,2,3] is solved by instantiating variables p and s to lists so that their concatenation yields the list [1,2,3] Thus, we can define a constraint which is satisfied if p is a prefix of the list xs as follows: prefix(p,xs) let s=s in conc(p,s) xs In order to show an example for higher order ....

[Article contains additional citation context not shown here]

E. Albert, S. Antoy, and G. Vidal. Measuring the E#ectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of the 10th Int'l Workshop on Logic-based Program Synthesis and Transformation (LOPSTR'00), pages 103--124. Springer LNCS 2042.

....computation [15] and positive supercompilation [30] the closest to our framework [4, 6] As a starting point, we extend the standard semantics for functional logic programs in at form the LNT calculus [21] with abstract costs. Our abstract costs are based on the cost criteria introduced in [1, 11] to measure the cost of functional logic computations: the number of steps, the number of pattern matching operations, and the number of applications. The resulting cost augmented semantics is similar to the one introduced in [5] for performing sourcelevel abstract pro ling. Nevertheless, here we ....

....centers but attribute all costs to a single cost center associated to the entire computation. Thus, our calculus is notably simpli ed. Basically, the enhanced semantics mimics the LNT calculus but additionally computes the abstract costs attributed to a particular computation. In contrast to [1, 11], we de ne our cost semantics for at programs. This makes our approach more practically applicable, since actual implementations of functional logic languages use a at representation as an intermediate language and, thus, realistic costs should be gathered at this level. Narrowing driven PE ....

[Article contains additional citation context not shown here]

E. Albert, S. Antoy, and G. Vidal. Measuring the Eectiveness of Partial Evaluation in Functional Logic Languages. In Proc. of the 10th Int'l Workshop on Logic-based Program Synthesis and Transformation (LOPSTR'00), pages 103-124. Springer LNCS

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