Lee Naish. Negation and Control in Prolog. Springer-Verlag, 1986. Lecture Notes in Computer Science, vol. 238.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....of concurrent and constraint logic programming. AKL encapsulates search by using nested computation spaces. A computation space is a constraint store with its associated goals. Search is done by allowing procedures to be defined by 4 The non declarative aspect has received some attention, e.g. [25, 28, 3]. 11 Kind of entity Protocol Entity Stateless Replication Eager record, procedure, class Lazy object record Single assignment Elimination Eager logic variable Lazy logic variable Stateful Localization Mobile cell, object state Stationary port, thread Table 3: Semantics of Distributed Oz ....

Lee Naish. Negation and Control in Prolog. Springer-Verlag, 1986. Lecture Notes in Computer Science, vol. 238.

....is not, but is possibly the most interesting since it is both declarative and computable. This semantics extends to CLP programs, as shown in [54] General implementations of these semantics are not well advanced. Negationas failure 3 can be implemented in the same way as in logic programming [43], but to make this rule sound execution of such sub goals should be delayed until the accumulated constraints determine a single value for each argument of q. Kunen s semantics can be implemented by constructive negation [54] Implementations of negation intending to produce answers need to have ....

L. Naish, Negation and Control in PROLOG, Lecture Notes in Computer Science 238, Springer-Verlag, 1986.

....substitution as its idempotent mgu. For example, the inequation x 6= y is effectively ground in the system with equations fz = f(y) x = ag. Theorem 3 A redundancy free system of equations and inequations has an mgss iff each inequation is effectively ground. In some applications, for example [33, 19], universally quantified variables appear naturally in the inequations. For example, consider the logic program P (g(x) Q(g(y) not P (y) R(g(z) not Q(z) The answer substitutions for R(w) are just those assignments for w that satisfy the system w = g(g(g(x) or the system w = g(z) and ....

....does not introduce inequalities. The above method has the advantage that negated goals are not executed in one indivisible computation, but can generate substitutions and negated substitution formulas during execution. When combined with a sophisticated execution mechanism, as in MUand NU Prolog [33], this allows co routining between positive and negative goals. The runtime form of constructive negation does not have this advantage, but applies to many more programs. The simple form of the runtime method [1] given a negative goal :A, executes A; if A has a finite SLD tree with answer ....

L. Naish, Negation and Control in Prolog, Lecture Notes in Computer Science 238, SpringerVerlag, 1986.

....level of e ciency, because the search space generated by the nondeterministic evaluation of a program is explored without taking into account any information derived from the program. Much work has been done in the direction of improving the control strategy of logic languages (see, for instance, [26, 110]) 20 We consider here a transformation technique, called compiling control [25] which follows a di erent approach. Instead of enhancing the naive Prolog evaluator using a clever (and often more complex) control strategy, we transform the given program so that the derived program behaves using ....

L. Naish. Negation and Control in Prolog. Lecture Notes in Computer Science 238. SpringerVerlag, 1985. 84

....which only gives rise to one ris transition. A second approach is to allow the programmer to annotate parts of the program (atoms, predicates, clauses, to provide a more flexible computation rule that is, nonetheless, programmed. This approach was pioneered in Prolog II [59] and MU Prolog [195]. The automatic annotation of programs [194] brings this approach closer to the first. A third approach is to introduce constructs from concurrent logic programming into the language. There are basically two varieties of this approach: guarded rules and guarded atoms. The former introduces a ....

L. Naish, Negation and Control in PROLOG, Lecture Notes in Computer Science 238, Springer-Verlag, 1986.

....generation of huge and hard to compile C programs. We believe that such work could benefit from well established work on portable C compilers. 4. 1 New execution models From the very beginnings of logic programming there has been a desire to obtain good execution models [Kow79] PP79] Col86] Nai85] More recently the need for such models has been made even clearer by the new goal of exploring parallelism in logic programs. One such important model is the Warren s Basic Andorra Model [War88] used in the the Andorra I system [SCWY91b, SCWY91c] In this model determinate goals (that is, ....

Lee Naish. Negation and Control in Prolog. Lecture notes in Computer Science 238. Springer--Verlag, 1985.

....not clear how to automatically identify the right directional types for a predicate. Another direction for future work is to explore applying these techniques to other problems in logic programming that depend on type information. For example, the techniques for analyzing control in sequential [Nai86], concurrent [Sha89] and parallel [Gre87] logic programming languages rely on knowledge about predicate types. Acknowledgements We would like to thank Saumya Debray, Nevin Heintze, Uday Reddy, Moshe Vardi, and Ed Wimmers for discussions and comments on some of the ideas presented in this paper. ....

Lee Naish. Negation and control in Prolog. Lecture Notes in Computer Science 238. SpringerVerlag, 1986.

....because the search space generated by the nondeterministic evaluation of a program is explored without using any information about the program. Much work has been done in the direction of improving the control strategy of logic languages (see, for instance, Bruynooghe and Pereira, 1984; Naish, 1985 ] We consider here a transformation technique, called compiling control [ Bruynooghe et al. 1989 ] which follows a different approach. Instead of enhancing the naive Prolog evaluator using a clever (and often more complex) control strategy, we transform the given program so that the ....

....statement. Prolog program transformations based on the insertion of cuts are reported in [ Sawamura and Takeshima, 1985; Debray and Warren, 1989; Deville, 1990 ] Other techniques which introduce annotations for the evaluator are related to the automatic generation of delay declarations [ Naish, 1985; Wiggins, 1992 ] which procrastinate calls to predicates until they are suitably instantiated. A final form of annotation technique which has been used for improving program efficiency is the so called memoing [ Michie, 1968 ] Results of previous computations are stored in a table together ....

L. Naish. Negation and Control in Prolog. Lecture Notes in Computer Science 238. Springer-Verlag, 1985.

....Prolog were not developed to run on parallel architectures. However, they provide parallelism of which a parallel implementation can take advantage. In particular, Prolog supports an architecture neutral declarative programming style especially if the language is extended by wait declarations [24] or parallel execution facilities as in Andorra [12] Inter process communication is not supported in Andorra: parallelism is exploited only implicitly for efficiency reasons. Ada provides parallel language aspects (rendezvous) and integrated communication primitives to coordinate sequential ....

Naish, L. Negation and Control in Prolog, vol. 238 of Lecture Notes in Computer Science. Springer Verlag, Berlin, 1986.

....unifies with a nogood. We cannot prune hypotheses because an instance is inconsistent. However, when computation progresses, we may substitute a value for a variable that makes the partial explanation inconsistent. This problem is similar to the problem of delaying negation as failure derivations [ Naish, 1986 ] and of delaying consistency checking in Theorist [ Poole, 1991a ] We would like to notice such inconsistencies as soon as possible. In the algorithm of Figure 1 we check for inconsistency each time a partial explanation is taken off the queue. There are cases where we do not have to check ....

L. Naish. Negation and Control in Prolog. Lecture Notes in Computer Science 238. Springer Verlag, 1986.

....versions of Concurrent Prolog and GHC, called FCP and FGHC respectively, were the focus of much work [8, 22] The KL1 language, derived from FGHC, was implemented in the high performance KLIC system. This system runs on sequential, 3 The non declarative aspect has received some attention, e.g. [18, 19, 7]. Kind of entity Protocol Entity Stateless Replication Eager record, procedure, class Lazy object record Single assignment Binding Eager logic variable Lazy logic variable Stateful Localization Mobile cell, object state Stationary port, thread Table 5: Semantics of Distributed Oz parallel, and ....

Lee Naish. Negation and Control in Prolog. Springer-Verlag, 1986. Lecture Notes in Computer Science, vol 238.

....of a logic program very often depends on the control strategy. Thus, in order to achieve high performances, the programmer, instead of relying on the evaluation strategy provided by the system, may define his own control strategy. This can be done, for instance, via modes or delay declarations [33] based on the instantiation patterns of the goals during execution. However, one may avoid the difficulty of dealing with those declarations at run time by using the compiling control technique as we now indicate. Let S left be the familiar Prolog control strategy, which selects the literals in ....

L. Naish. Negation and Control in Prolog. Lecture Notes in Computer Science

....and Prolog were not developed to run on parallel architectures. However, they provide parallelism of which a parallel implementation can take advantage. In particular, Prolog supports an architecture neutral declarative programming style especially if the language is extended by wait declarations [22] and concurrent execution facilities. Ada provides parallel language aspects (rendezvous) and integrated communication primitives to coordinate sequential processes. The parallel conception of occam s synchronous channels is more pronounced than that of Ada. The languages at the serializability ....

L. Naish: Negation and Control in Prolog. Springer, Lecture Notes in Computer Science, Vol. 238, Berlin, 1986.

....unifies with a nogood. We cannot prune hypotheses because an instance is inconsistent. However, when computation progresses, we may substitute a value for a variable that makes the partial explanation inconsistent. This problem is similar to the problem of delaying negation as failure derivations [13], and of delaying consistency checking in Theorist [20] We would like to notice such inconsistencies as soon as possible. In the algorithm of Figure 1 we check for inconsistency each time a partial explanation is taken off the queue. There are cases where we do not have to check this explicitly, ....

L. Naish. Negation and Control in Prolog. Lecture Notes in Computer Science 238. Springer Verlag, 1986.

....in the B Prolog system. For several benchmark programs where delay is used, the experimental results show that B Prolog is significantly faster and sometimes consumes much less total space than SICStus, a WAM based Prolog system. 1 Introduction Delay has become a mechanism common to logic [3, 5, 12], functional logic [1, 8] constraint logic [9, 16] and concurrent logic [13, 14] programming languages. It relaxes the strict left to right computation rule adopted in Prolog and enables the execution of some predicate calls to be delayed until some variables in them are instantiated. Delay is ....

....constraints are delayed until they are solvable. Delay mechanism has been found useful in a wide range of applications. It can be used to describe perpetual processes and deal with infinite data structures [14] It allows a generate and test program to be written as a test and generate program [12]. The latter is usually much faster than the former because tests can be performed as soon as they become testable and failure can be detected early. Delay mechanism also makes it possible to implement sound versions of negation and arithmetic [12] In addition, it has been used to implement ....

[Article contains additional citation context not shown here]

Naish, L.: Negation and Control in Prolog, Lecture Note in Computer Science, 238, 1985.

....This system runs on sequential, parallel, and distributed machines [14] A number of implementation techniques in the current Distributed Oz system have been borrowed from KLIC, notably the distributed garbage collection algorithm. 5 The non declarative aspect has received some attention, e.g. [36, 39, 3]. Kind of entity Protocol Entity Stateless Replication Eager record, procedure, class Lazy object record Single assignment Binding Eager logic variable Lazy logic variable Stateful Localization Mobile cell, object state Stationary port, thread Table 3: Semantics of Distributed Oz An important ....

Lee Naish. Negation and Control in Prolog. Springer-Verlag, 1986. Lecture Notes in Computer Science, vol. 238.

....course, the implementation is still sound. 7 Related Work The declarative semantics is based on completion semantics developed for negation by Kunen [12] for logic programs, and extended to constraint logic programs by Stuckey [19] Our operational semantics is related to that proposed by Naish [17] for negation and aggregation, in that optimization subgoals must delay until their global variables have a fixed value. Using the current best optimum to prune the search space is related to the operational semantics of optimization in 2LP [16] and CHIP s minimize(G,M) predicate [3] Our proofs ....

Naish, L. Negation and Control in Prolog. Lecture Notes in Computer Science 238. Springer Verlag, New York, 1986.

No context found.

L. Naish. Negation and Control in Prolog. Lecture Notes in Computer Science , n o 238, Springer-Verlag 1986.

No context found.

L. Naish. Negation and Control in Prolog. Lecture Notes in Computer Science , n o 238, Springer-Verlag 1986.

No context found.

Naish, L. Negation and Control in Prolog. Springer-Verlag, Lecture Notes in Computer Science 238, Berlin, 1986.

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