G. Levi and G. Sardu. Partial Evaluation of Metaprograms in a "Multiple Worlds" Logic Language. New Generation Computing, 6(2,3):227--247, 1988.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....New M that is more efficient than M . We will now compare M and New M as far as memory occupation and execution times are concerned. 5. 1 Memory occupation When considering memory occupation, a typical risk of applying standard partial evaluation techniques is represented by code explosion [2, 9, 29]. In our case, as we shall discuss below, the 32 partial evaluation strategy employed ensures that the size of the specialised program New M is approximately the same as the size of the initial program M . Let us first make some observations on the relation between the number of clauses of M ....

....our specialiser with Mixtus or SP concerns declarativeness. Namely, the declarative style of our program specialiser using Godel ground representation of object programs is not supported by Mixtus and SP, which employ Prolog s non declarative features. 6 Related work Levi and Sardu [29] made one of the first experiments of applying partial evaluation techniques to multiple logic program settings. They describe the realisation of a partial evaluator for a knowledge base management system, where separate Prolog programs can be combined by means of inheritance operators. Their ....

[Article contains additional citation context not shown here]

G. Levi and G. Sardu. Partial Evaluation of Metaprograms in a "Multiple Worlds" Logic Language. New Generation Computing, 6(2,3):227--247, 1988.

....be included under this heading. Procedures for querying, updating, checking integrity constraints, and similar tasks are thus meta level procedures. The potential of using partial evaluation to optimize expert system query interpreters with respect to fixed bases of knowledge was identified by Levi and Sardu (1988) and by Sterling and Beer (1989) Enhanced Language Interpreters. In logic programming, interpreters for tracing computations, spying, timing and so on are sometimes written as enhanced versions of a standard or vanilla interpreter (see the ProofTree interpreter in Figure 4) That is, the ....

Levi, G. & Sardu, G. (1988), `Partial evaluation of metaprograms in a "multiple worlds" logic language', New Generation Computing 6(2,3), 227-- 247.

....yield a new, specialised meta program M s that is equivalent to M w.r.t. queries of the form Demo(E,G) for any object query G. We will indeed show how the general results described in Sect. 3 can be suitably applied to establish such an equivalence result. As pointed out for instance in [2, 24], it is crucial to develop some strategy to control unfolding. For instance, in our case, the unfolding of predicate Demo hardly ever increases the degree of specialisation of the meta interpreter M w.r.t. a query of the form Demo(E,x) Indeed, since the second argument of Demo in the initial ....

....New M that is more efficient than M . We will now compare M and New M as far as memory occupation and execution times are concerned. 5. 1 Memory occupation When considering memory occupation, a typical risk of applying standard partial evaluation techniques is represented by code explosion [2, 9, 24]. In our case, as we shall discuss below, the partial evaluation strategy employed ensures that the size of the specialised program New M is approximately the same as the size of the initial program M . Let us first make some observations on the relation between the number of clauses of M ....

[Article contains additional citation context not shown here]

G. Levi and G. Sardu. Partial Evaluation of Metaprograms in a "Multiple Worlds" Logic Language. New Generation Computing, 6(2,3):227--247, 1988.

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