Gert Smolka and Ralf Treinen (ed.). DFKI Oz documentation series. Deutsches Forschungszentrum fur Kunstliche Intelligenz, Stuhlsatzenhausweg 3, D--66123 Saarbrucken, Germany, 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....and fair threads. Contents 1 Introduction 3 2 An Informal Computation Model 4 3 An Example: Mapping Lists 7 4 The Constraint Store 8 5 Introducing Amoz 9 6 Threads and Matching 11 7 Local Computation Spaces 13 8 First class Procedures 17 1 Introduction Oz is a concurrent constraint language [20, 19, 17, 6, 22] providing for functional, object oriented, and constraint programming. It has a simple yet powerful computation model [19, 20] which extends the concurrent constraint model [10, 16] by first class procedures, deep guards, concurrent state, and encapsulated search. DFKI Oz [11] is an interactive ....

Gert Smolka and Ralf Treinen (ed.). DFKI Oz documentation series. Deutsches Forschungszentrum fur Kunstliche Intelligenz, Stuhlsatzenhausweg 3, D--66123 Saarbrucken, Germany, 1994.

.... a useful exension to the Herbrand constraint theory, found in many logic programming languages, since feature trees are more general than terms [AKSP91] Records are restricted, compared to features trees, in that 1 they have fixed sort and arity but they can still provide for a constraint theory [ST92]. When using Herbrand terms for building data structures, there is always the problem of accessing a certain subterm of a given term. For every such access a special selector must be created. The programmer must also remember what subterm to access since the term itself does not contain any ....

....searched for and a link is set to this entry. 6 Implementation of records in AGENTS. Herbrand trees (the terms of Prolog) herafter known as trees, can be viewed as records where the name of the structure is the sort and the features are consecutive numbers 1: N, N being the arity of the tree [ST92]. The value for feature n correspond to argument n in the tree. Atoms and numbers are considered to be records without features. Ex: The tree X = foo(a,bar(b,1) Y) can be described by the following constraints: foo X The sort. X [1,2,3] The features. X 1 a, X 2 bar(b,1) X 3 Y The values of ....

G. Smolka and R. Treinen. Records for logic programming. Technical report, Deutsches Forschungszentrum fur Kunstliche Intelligenz GMbH, August 1992.

.... programming platform ECL i PS e (see appendix 1) At the moment, there exist two sequential implementations, one prototype in LISP [Her93] and one fully developed CHRs library in ECL i PS e [B 95] At DFKI Saarbrucken, an implementation of CHRsin the concurrent object oriented language OZ [SmTr94] is on the way. The LISP implementation does not provide for simpagation rules, but offers some interesting extensions. First, rules can be given priorities (encoded as integers) Second, indeterminism is introduced by disjunction in rule bodies. This extension also allows to express Prolog ....

Gert Smolka and Ralf Treinen (ed.), DFKI Oz Documentation Series, Deutsches Forschungszentrum fur Kunstliche Intelligenz, Stuhlsatzenhausweg 3, D--66123 Saarbrucken, Germany, 1994, available via WWW from http://ps-www.dfki.uni-sb.de/oz/.

....been further developed for use in the CCP type languages Life (an early description in [AKL88] AKP93b] and Oz [HSW93] also see [AKPS92, AKP93a] Another early precursor are feature descriptions found originally in linguistics. Mukai has described them using a record algebra model [Muk91] In [ST92] an algorithm for constraint simplification in feature trees is presented. The algorithm provides for incremental entailment and disentailment tests of simple constraints, i.e. atomic feature constraints closed under conjunction and existential quantification. It is also discussed how ....

Gert Smolka and Ralf Treinen. Records for Logic Programming. Research Report RR-93-23, Deutsches Forschungszentrum fur Kunstliche Intelligenz GmbH, (DFKI), Saabrucken, Germany, Aug 1992.

....away from the full details of the network and differences in data representation. Furthermore the access model offers a common information model and a uniform language for access and modification of the spaces. The platform we consider is the mOzArt platform [30] for distributed programming in Oz [64, 65, 66]. By providing a communication layer for sharing of distributed objects the mOzArt platform will give a unique support for the design of models such as the one discussed above. We will base the design of this layer on the emerging standard FIPA for agent interaction [23] 3 The Railway Scheduling ....

Smolka, G., Treinen, R. eds. DFKI Oz Documentation Series. Deutsches Forschungszentrum fur Kunstliche Intelligenz GmbH, Stuhlsatzenhausweg 3, 66123 Saarbrucken, Germany 1995.

....were not sufficient. Hence, techniques from OR were incorporated into constraint systems, pioneered by CHIP [4] Changes of the problem formulation can be modeled by stating different constraints. But CHIP lacks flexibility to allow the user to invent new kinds of constraints. We propose Oz [11, 12] as a platform to integrate algorithms from OR to achieve an amalgamation of a high level constraint language with efficient OR techniques. Oz is a concurrent constraint language providing for functional, object oriented, and constraint programming. The unique advantages of Oz, which can be ....

G. Smolka and R. Treinen, editors. DFKI Oz Documentation Series. Deutsches Forschungszentrum fur Kunstliche Intelligenz GmbH, Stuhlsatzenhausweg 3, 66123 Saarbrucken, Germany, 1995.

....scheduling compiler. While the first implementation of the Scheduler deals with one class of scheduling problems only (see Section 3) it can be extended to handle further classes by the dynamic extension to different scheduling compilers (see Section 4. 2) As an implementation language we use Oz [22, 23, 13]. Oz is a concurrent constraint language providing for functional, object oriented, and constraint programming, which also features programmable search. Thus, Oz appears to be a promising candidate for implementing a workbench like this Scheduler. Through the use of the constraint interface of ....

....the propagator to strengthen the store to X;Y 2 f1; 3g and Z = 5. Imposing X = 3 lets the propagator narrow the domain of Y to one. A tool for displaying arbitrary data structures and the current domains of variables is the Oz Browser. Changes in the domains cause an update of the display [23]. A solution to a set of finite domain constraints is a mapping from variables to nonnegative integers. To obtain a solution for a set of constraints S we usually have to choose a (not necessarily basic) constraint C and solve both S[fCg and S [f:Cg; we distribute S with C at the current ....

[Article contains additional citation context not shown here]

G. Smolka and R. Treinen, editors. DFKI Oz Documentation Series. Deutsches Forschungszentrum fur Kunstliche Intelligenz GmbH, Stuhlsatzenhausweg 3, 66123 Saarbrucken, Germany, 1995.

....y = z. Hence by Proposition 4. 2 ) j ) j=j ) j= 5) The last implication is justified by V( C(j ) hence V( j ) Now, the claim follows by (3) from (4) and (5) 2 A proof of a more general lemma (in the context of feature constraints) has been given in [ST92b]. Before we state the entailment theorem we have to introduce some more notation. We call X directed if contains no equation x = y with x 62 X and y 2 X. For a constraint OE we define OE only variables from X, and OE only variables alien to X. Since every constraint is either ....

....the right hand side. The considerations for the memo 1 rule are similar. 2 9 Conclusion and Further Work We have presented a rule based algorithm which allows for satisfiability and entailment tests of equational and membership constraints. The development of an abstract machine in the style of [ST92b] and the calculation of precise complexity bounds is up to further research. The constraint system presented here can possibly be extended in various directions. One immediate question is the decidability of the first order theory of a deterministic equation system with maximal fixpoint solution; ....

Gert Smolka and Ralf Treinen. Records for logic programming. Research Report RR-92-23, Deutsches Forschungszentrum fur Kunstliche Intelligenz, Stuhlsatzenhausweg 3, D-W-6600 Saarbrucken, Germany, August 1992.

....take a more elementary view. We use the language of so called features, which are well known for a long time in computational linguistics and knowledge representation (see [12] for a survey) There are different possible choices of a feature language. A first language FT has been established in [3] and [1] The language of FT consists of unary label predicates, which express that the root node of a tree has a certain label, and binary feature predicates, which serve as the partial selector functions for trees. For instance, we can translate a Herbrand formula x = f(x 1 ; xn ) into ....

....of the algebra of rational trees, using the language of Herbrand, have been given independently in [5] for the case of a finite signature, and in [11] for both the case of a finite and an infinite signature. A complete axiomatization of rational trees in the language of FT has been given in [3], and a complete axiomatization of rational trees in the language of CFT in [2] In both cases, a quantifier elimination method has been used with a similar overall structure than [11] Both methods for proving the completeness of CFT have their merits. The quantifier elimination used in [2] ....

[Article contains additional citation context not shown here]

Rolf Backofen and Gert Smolka. A complete and recursive feature theory. Research Report RR-92-30, Deutsches Forschungszentrum fur Kunstliche Intelligenz, Stuhlsatzenhausweg 3, D 66123 Saarbrucken, Germany, September 1992. Short version as [4].

....of lazy and eager calculus. We should remark that the ffi calculus is unable to model optimal reduction [GAL, Lam90, Kat90] since it prohibits reduction in abstractions. The ffi calculus is one of several calculi [Smo93] describing aspects of the multi paradigm programming language Oz [Smo94]. All those calculi use relational settings as does the ffi calculus. Oz has been developed in parallel with those calculi allowing for specifications and discussions of language features. Since Oz is implemented, the ffi calculus is. The converse does not hold. Oz provides for many other ....

Gert Smolka et al. The Oz Handbook. Research Report RR-94-??, Deutsches Forschungszentrum fur Kunstliche Intelligenz, Stuhlsatzenhausweg 3, D-66123 Saarbrucken, Germany, 1994. Available through anonymous ftp from duck.dfki.uni-sb.de.

.... j=j 89C( j 0 ) 0 j 0 ) j= 89C(j 0 ) 0 j 0 ) 5) The last implication is justified by V( C(j 0 ) hence V( j 0 ) C(j 0 ) Now, the claim follows by (3) from (4) and (5) 2 A proof of a more general lemma (in the context of feature constraints) has been given in [ST92b]. Before we state the entailment theorem we have to introduce some more notation. We call X directed if contains no equation x : y with x 62 X and y 2 X. For a constraint OE we define OE X to be the subset of atomic constraints which constrain only variables from X, and OE GammaX to ....

....the right hand side. The considerations for the memo 1 rule are similar. 2 9 Conclusion and Further Work We have presented a rule based algorithm which allows for satisfiability and entailment tests of equational and membership constraints. The development of an abstract machine in the style of [ST92b] and the calculation of precise complexity bounds is up to further research. The constraint system presented here can possibly be extended in various directions. One immediate question is the decidability of the first order theory of a deterministic equation system with maximal fixpoint solution; ....

Gert Smolka and Ralf Treinen. Records for logic programming. Research Report RR-92-23, Deutsches Forschungszentrum fur Kunstliche Intelligenz, Stuhlsatzenhausweg 3, D-W-6600 Saarbrucken, Germany, August 1992.

....systems FT [3, 5] and CFT [15] The latter constraint system subsumes Colmerauer s classical rational tree constraint system [6] but provides for finer grained and more expressive constraints. An efficient implementation of tests for satisfiability and entailment in CFT has been given in [16]. In fact, satisfiability of CFT constraints can be tested in at most quadratic time, and for a mildly restricted case in quasi linear time. CFT is the theoretical base for the constraint system of Oz [14] CFT s standard model consists of so called feature trees, that is possibly infinite trees ....

G. Smolka and R. Treinen. Records for logic programming. Research Report RR-92-23, Deutsches Forschungszentrum fur Kunstliche Intelligenz, Stuhlsatzenhausweg 3, D-W-6600 Saarbrucken, Germany, Aug. 1992.

No context found.

R. Backofen and G. Smolka. A complete and recursive feature theory. Research Report RR-92-30, Deutsches Forschungszentrum fur Kunstliche Intelligenz, Stuhlsatzenhausweg 3, D 66123 Saarbrucken, Germany, Sept. 1992.

No context found.

Smolka, G., Treinen, R. eds. DFKI Oz Documentation Series. Deutsches Forschungszentrum fur Kunstliche Intelligenz GmbH, Stuhlsatzenhausweg 3, 66123 Saarbrucken, Germany 1995.

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