R.E. Fikes. REF-ARF: A system for solving problems stated as procedures. Artificial Intelligence, 1:27--120, 1970.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....The Huffman Clowes Waltz scene labeling problem is a Finite CSP (FCSP) since the variable domains are discrete and finite. Our complexity results were for FCSPs. One of the key insights of arc consistency for FCSPs can be found in Fikes paper in the very first issue of Artificial Intelligence [6]; in particular, if a value, for one problem variable is inconsistent with all values for some other problem variable, then will never participate in a complete solution to the problem and can be eliminated from all further consideration. The obvious algorithm for removing all such ....

.... each level trades increased representational power for increased complexity [7] One of the practical consequences of our results was that the designers and implementers of constraint based programming languages could feel comfortable including consistency algorithms as primitives in the language [6,10]. Ideally, a language primitive should require constant time; but, failing that, it is comforting to know that it will terminate in linear time. The constraint logic programming language CHIP [20] was the first to exploit this potential fully by providing an arc consistency based inference ....

Fikes, R. E. REF-ARF: a system for solving problems stated as procedures. Artificial Intelligence 1 (1970), 27--120.

....automatically constructed and used as counterexamples in actual proofs [4] They have discussed that examples can also be used positively examples are used in conjecturing lemmas which can be proved and which, in turn, are used in establishing the proof of the main theorem. 7. 4 REF ARF REF ARF [16] was designed as a problem solving system such as GPS; its problem solving method is constraint satisfaction. REF is a language for stating problems, which is an extension of an ALGOL like syntax. ARF, the REF interpreter, translates the REF problem statement into the form of contexts , internal ....

Fikes, R. E. "REF-ARF: A System for Solving Problems Stated as Procedures". Artificial Intelligence 1 (1970), 27-120.

....Some other related systems, EL ARS [238] and SYN [142] used the constraint solver MACSYMA [186] to avoid the restrictions of local propagation. It was noted at this period [241] that there was a conceptual correspondence between the constraint techniques and logic programming. The REF ARF system [89] was also designed for problem solving. One component, REF, was essentially a procedural language, but with nondeterminism because of constraints used in conditional statements. The constraints are static. They are in fact linear integer constraints, and all variables are bounded above and below. ....

R.E. Fikes, REF-ARF: A system for solving problems stated as procedures, Artificial Intelligence 1, 27--120, 1970.

....local propagation, it was used in applications for electrical circuit analysis and synthesis. EL ARS 2 [35] and SYN [11] was related systems that used the constraint solver MACSYMA [32] to avoid the restrictions of local propagation. An early system designed for problem solving was the REF ARF [15] system. REF is a procedural language with nondeterminism because of static constraints used in conditional statements, ARF is the constraint solver using backtracking. During this time a lot of work was also done in modeling combinatorial problems as Constraint Satisfaction Problems (CSP s) and ....

R. E. Fikes. REF-ARF: a system for solving problems stated as procedures. Artificial Intelligence, 1:27--120, 1970.

....in the program as a set of facts, or even as a set of rules [Van89] We shall often refer to constraints in the CSP framework as propagation constraints . One prerequisite for applying CSP techniques is that problem variables should have an associated domain of possible values. Traditionally [Fik70, Mon74, Mac77, HE80] this is a finite domain, though more recently continuous intervals have been studied [Dav87] Up to now, constraint logic programming systems based on the CSP paradigm (eg CHIP [DVS 88] have therefore had to include domain declarations for the problem variables. This is ....

....88] have therefore had to include domain declarations for the problem variables. This is one restriction we will relax using generalised propagation. 4 It should be noted that constraint solving over a computation domain is replaced in this paradigm by constraint propagation over value domains [Fik70, Mon74, Mac77, HE80] Informally constraint propagation operates by looking ahead at yet unsolved goals to see what locally consistent valuations there remain for individual problem variables. In the CSP framework there is no guarantee that, after a complete propagation sequence, the propagation ....

R.E. Fikes. REF-ARF: A system for solving problems stated as procedures. Artificial Intelligence, 1:27--120, 1970.

....pine.ece.utexas.edu in a pine home pine ftp pub parprosys. In the measurements presented in section 6.1, instances of the CSTP appear as south, south2, moun and moun2. 4. 2 Confusion of Patents Problem We constructed a solution for the formulation of the Confusion of Patents Problem presented in [19, 40] The problem presents five patents, five inventors, five cities, and ten constraints. Using these constraints we must decide who invented what and where. In our solution, all 125 possible combinations and 10 constraints are present in the initial database; 67 productions use the constraints to ....

R. E. Fikes, REF-ARF: A system for solving problems stated as procedures, Artificial Intelligence, 1 (1970), pp. 27--120.

....The Huffman Clowes Waltz scene labeling problem is a Finite CSP (FCSP) since the variable domains are discrete and finite. Our complexity results were for FCSPs. One of the key insights of arc consistency for FCSPs can be found in Fikes paper in the very first issue of Artificial Intelligence [6]; in particular, if a value, c, for one problem variable is inconsistent with all values for some other problem variable, then c will never participate in a complete solution to the problem and can be eliminated from all further consideration. The obvious algorithm for removing all such ....

.... each level trades increased representational power for increased complexity [7] One of the practical consequences of our results was that the designers and implementers of constraint based programming languages could feel comfortable including consistency algorithms as primitives in the language [6,10]. Ideally, a language primitive should require constant time; but, failing that, it is comforting to know that it will terminate in linear time. The constraint logic programming language CHIP [20] was the first to exploit this potential fully by providing an arc consistency based inference ....

Fikes, R. E. REF-ARF: a system for solving problems stated as procedures. Artificial Intelligence 1 (1970), 27--120.

....recently, Schemer, a non deterministic dialect of Lisp [110] was used as a front end to a dependency directed search engine. But perhaps the most comprehensive attempt at providing intelligent support for solving a range of search problems, and certainly one of the earliest, is the REF ARF system [41]. The REF language is a typical late 60 s imperative programming language, reminiscent of ALGOL, except for two special language constructs called select and condition which provided support for describing non deterministic algorithms. The select function takes two integer arguments, and ....

....the programmer. CHAPTER 5. NON DETERMINISTIC UPDATES IN P FDM 152 The results are impressive. For example, under REF ARF, the well known cryptarithmetic problem SEND MORE=MONEY reduces to a search space of four solutions after constraint manipulation, and is solved in only two backtracking steps [41]. This compares very favourably with the recent CHIP system which requires one backtracking step after constraint manipulation [53] All these languages extend some deterministic language with new constructs for describing the non deterministic assignment of values to variables, and for describing ....

R.E. Fikes. REF-ARF: a System for Solving Problems Stated as Procedures. Artificial Intelligence, 1:27--120, 1970.

....) C. Castro ) ff 0 2 SolD (x j 2 D x j c j (x j ) c k (ff(x i ) x j ) A network of constraints is arc consistent if all its arcs are consistent. The rst three algorithms developed to achieve arc consistency are based on the following basic operation rst proposed by Fikes [8]: Given two variables x i and x j , both of which are node consistent, and the constraint c (x i ; x j ) if ff 2 SolD (x i 2 D x i ) and there is no ff 0 2 SolD (x j 2 D x j c (ff(x i ) x j ) then ff(x i ) has to be deleted from D x i . When that has been done for each ff ....

R. E. Fikes. REF-ARF: A System for Solving Problems Stated as Procedures. Artiøcial Intelligence, 1:27120, 1970.

....class of CSPs in this framework and show how to efficiently combine and generate new heuristics. Finally, we present some experimental results. 1 Introduction Since in general combinatorial problems are NP hard, current practical approaches for solving them employ heuristics. For CSPs (see [ Fikes, 1970 ] Freuder, 1978 ] finding a solution means constructing an assignment of values to variables such that all constraints are satisfied. A solution is constructed by stepwise extension of partial solutions including backtracking whenever necessary. Basic heuristics for CSPs are algorithms to ....

....are presented in the next section. Finally, Section 6 summarizes the paper. 2 Definitions and Preliminaries 2. 1 Constraint Satisfaction Problems A constraint is usually defined in the CSP notation as a set of allowed combinations of values for some subset of a finite set of variables (cf. Fikes, 1970 ] Solving a CSP implies searching for combinations of values for all variables satisfying all constraints of this CSP. We now give a formal definition: Definition 2.1 (constraint satisfaction problem) A constraint satisfaction problem P is a triple (V ; D; C) where ffl V is a finite ....

R. E. Fikes. REF-ARF: A System for Solving Problems Stated as Procedures. Artificial Intelligence, 1, 1970.

....during the computation (which distinguishes constraints from ordinary built in predicates [Mah87] This check must, in theory, be effective. 1. 2 CSP in Logic Programming There is another, very different, CLP paradigm which is based on constraint satisfaction techniques dating back to 1965 [GB65, Fik70, Mon74] In the constraint satisfaction problem (CSP) paradigm the constraints are problem specific, and defined by sets of tuples. When CSP is embedded into logic programming, a constraint can be defined in the program as a set of facts, or even as a set of rules [Van89] We shall often refer to ....

....use of constraints [Gal85] This is provided by techniques developed for solving constraint satisfaction problems. It should be noted that constraint solving over a computation domain, as described in section 1. 1 above, is replaced in this paradigm by constraint propagation over value domains [Fik70, Mon74, Mac77, HE80] Informally constraint propagation operates by looking ahead at yet unsolved goals to see what locally consistent valuations there remain for individual problem variables. In the CSP framework there is no guarantee that, after a complete propagation sequence, the propagation ....

R.E. Fikes. REF-ARF: A system for solving problems stated as procedures. Artificial Intelligence, 1:27--120, 1970.

....original problem would completely eliminate the need for search, but as mentioned before, this is usually more expensive than simple backtracking. A number of algorithms for solving CSPs have been investigated by various researcher that essentially fit the above format (Haralick 1980; Nadel 1988; Fikes 1970; Gaschnig 1974; Ullman 1976; Haralick 1978; McGregor 1979; Dechter 1989a) In particular, Nadel (Nadel 1988) empirically compared the performance of the following algorithms: Generate and Test (GT) Simple Backtracking (BT) Forward Checking (FC) Partial Lookahead (PL) Full Lookahead (FL) ....

Fikes, R. E. 1970. Ref-Arf: A System for Solving Problems Stated as Procedures.

....C 8 15 C 1 # WMEs in initial database C (2 2 c 2 c 3) C (2 2 c 2 c 3) Table 2: Static measures for the CTSP benchmark according to C and c 5.1. 3 Confusion of Patents Problem We constructed a solution for the formulation of the Confusion of Patents Problem presented in [10, 22] The problem presents five patents, five inventors, five cities, and ten constraints. Using these constraints we must decide who invented what and where. In our solution, all 125 possible combinations and 10 constraints are present in the initial database; 67 productions use the constraints to ....

R. E. Fikes. REF-ARF: A system for solving problems stated as procedures. Artificial Intelligence, 1(1):27--120, 1970.

....40 6.1 1.3 5 hotel 80 4.1 2.0 62 patents 86 5.2 1.2 4 south 91 4.7 2.8 40 south2 121 4.7 2.7 61 moun 211 4.7 2.8 88 moun2 301 4.7 2.7 151 waltz2 10 2.7 8.0 7 Table 1: Static Measures for Benchmarks Used. patents is our solution to the constraint Confusion of Patents Problem presented in [5]. Because this solution has only four different types of WMEs, most of the productions either change or test the same kinds of WME. As a consequence, productions have strong interdependency, resulting in a production system poorly suited for clustering. The main source of parallelism is the ....

R. E. Fikes. REF-ARF: A system for solving problems stated as procedures. Artificial Intelligence, 1(1):27--120, 1970.

....by examples, to explain the current success of constraint technology, and, by showing its benefits, to add to that success. In 1963 Sutherland introduced the Sketchpad system [118] a constraint language for graphical interaction. Other early constraint programming languages were Fikes Ref Arf [44], Lauriere s Alice [78] Sussmann s CONSTRAINTS [116] and Borning s ThingLab [10] These languages already offered the most important features of constraint programming: ffl declarative problem modelling ffl propagation of the effects of decisions ffl efficient search for feasible solutions ....

R. E. Fikes. REF--ARF: A system for solving problems stated as procedures. Artificial Intelligence, 1:27--120, 1970.

....40 6.1 1.3 5 hotel 80 4.1 2.0 62 patents 86 5.2 1.2 4 south 91 4.7 2.8 40 south2 121 4.7 2.7 61 moun 211 4.7 2.8 88 moun2 301 4.7 2.7 151 waltz2 10 2.7 8.0 7 Table 1: Static Measures for Benchmarks Used. patents is our solution to the constraint Confusion of Patents Problem presented in [6]. Because this solution has only four different types of WMEs, most of the productions either change or test the same kinds of WME. As a consequence, productions have strong interdependency, resulting in a production system poorly suited for clustering. The main source of parallelism is the ....

R. E. Fikes. REF-ARF: A system for solving problems stated as procedures. Artificial Intelligence, 1(1):27--120, 1970.

....to estimate performance in the multiple functional unit Rete network. south and south2 are CTSPs with four countries and ten cities per country; moun and moun2 are CTSPs with ten countries and fifteen cities per country. Our solution to the constraint Confusion of Patents Problem presented in [14, 24] is patents. Originally written by Steve Kuo at the University of Southern California, hotel is a production system that models the operation of a hotel. It is a relatively large and varied production system (80 productions, 65 WME types) with 17 non exclusive contexts. life is an implementation ....

R. E. Fikes. REF-ARF: A system for solving problems stated as procedures. Artificial Intelligence, 1(1):27--120, 1970.

No context found.

R.E. Fikes. REF-ARF: A system for solving problems stated as procedures. Artificial Intelligence, 1:27--120, 1970.

No context found.

Fikes, R. E. (1970). REF-ARF: A system for solving problems stated as procedures. Artificial Intelligence, 1, 27--120.

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