FUTAMURA, Y., AND NOGI, K. Generalized partial computation. In IFIP TC2 Workshop on Partial Evaluation and Mixed Computation (1987), D. Bjrner, A. P. Ershov, and N. D. Jones, Eds., North--Holland, pp. 133--151.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....empty) and subsumes dynamic slicing (because the prefix can be an entire input sequence) Quasi static slicing was motivated by work on partial evaluation. The concept of quasi static slicing is subsumed by conditioned slicing, which is the most general form of slicing. In 1987, Futamura and Nogi [28], introduced a generalization of partial evaluation, which generalizes the way in which partial evaluation specializes a program with respect to partial input information. Using general partial evaluation, the input information is captured by a predicate, just as the conditioned slicing criterion ....

Yoshihiko Futamura and Kenroku Nogi. Generalized partial computation. In D. Bjrner, Andrei P. Ershov, and Neil D. Jones, editors, IFIP TC2 Workshop on Partial Evaluation and Mixed Computation, pages 133--151. North--Holland, 1987.

....program point indexed by known values. In 1987, at the workshop on Partial Evaluation and Mixed Computation [9, 25] this very simple specialization strategy was challenged. Yoshihiko Futamura said that he had needed to design a stronger form of partial evaluation Generalized Partial Computation [26] to be able to specialize a naive quadratic time string matcher into Knuth, Morris, and Pratt s (KMP) linear time string matcher [39] He therefore asked the DIKU group whether polyvariant specialization was able to obtain the KMP. 1.1 Obtaining the KMP The naive matching algorithm tests ....

....been developed that keep a static trace of the dynamic prefix implicitly, making them able to pass the KMP test [49] i.e. to specialize the original quadratic string matching program into a KMPlike residual program. Such partial evaluators include Futamura s Generalized Partial Computation [26], Smith s partial evaluator for constraint logic programming languages [48] Queinnec and Ge#roy s intelligent backtracking [46] supercompilation [28, 29, 49, 50, 51] partial deduction [44] partial evaluators for functional logic programs [4, 40] and the composition of a memoizing interpreter ....

Yoshihiko Futamura and Kenroku Nogi. Generalized partial computation. In Bjrner et al. [9], pages 133--151.

....a quadratic string matcher that searches for the first occurrence of a pattern in a text, a partial evaluator specializes this string matcher with respect to a pattern and yields a residual program that traverses the text in linear time. The problem was first stated by Yoshihiko Futamura in 1987 [15] and since then, it has given rise to a variety of solutions [2, 3, 10, 13, 14, 16, 19, 25, 28, 29, 32, 33] For 15 years, however, it has also been pointed out that the traditional solution only solves half of the problem. Indeed, the Knuth Morris Pratt matcher first produces a next table in ....

Yoshihiko Futamura and Kenroku Nogi. Generalized partial computation. In Dines Bjrner, Andrei P. Ershov, and Neil D. Jones, editors, Partial Evaluation and Mixed Computation, pages 133--151. North-Holland, 1988.

....than the dynamic x. Negative knowledge may also be exploited. For instance, one knows that the dynamic x de nitely does not have the value of E 1 in the false branch of the conditional above. Improvements of these kinds were of great importance in [CD89] Some systems automate such improvements [Tur86, FN88]. 7.2.2 Dynamic choice of static values (car (if E 1 E 2 E 3 ) where E 1 is dynamic, but E 2 and E 3 static. The dynamic test will make the result of the conditional expression dynamic and hence no reduction of the car operation will take place. This is the classical problem of a conditional ....

Yoshihiko Futamura and Kenroku Nogi. Generalized partial computation. In Dines Bjrner, Andrei P. Ershov, and Neil D. Jones, editors, Partial Evaluation and Mixed Computation, pages 133-151, North-Holland, 1988.

....best observed through specializing a self interpreter (which amounts to testing whether a specializer is Jones optimal) A way to test the strength of a specializer is to see whether it can derive certain well known e#cient programs from naive and ine#cient programs. One of the most popular tests [10, 6, 29, 2] is to see whether the specializer generates, from a naive pattern matcher and a fixed pattern, an e#cient pattern matcher. What makes Jones optimality stand out in comparison to such tests is that while a Jones optimal specializer with static expression reduction is guaranteed to pass any of ....

Y. Futamura, K. Nogi. Generalized partial computation. In D. Bjrner, A. P. Ershov, N. D. Jones (eds.), Partial Evaluation and Mixed Computation, 133--151. North-Holland, 1988.

....by specialization with respect to the pattern. Indeed, this KMP test has become a popular benchmark for partial evaluators and related systems [21] The systems that pass the KMP test have the ability to infer information about the unknown input based on the form of enclosing conditional tests [6, 20, 21]. Such capability, however, goes beyond standard partial evaluation. Another approach is to rewrite a naive string matcher to make it more amenable to standard partial evaluation by augmenting the implementation with static data recording the results of tests on the dynamic data: A standard ....

....tests. In contrast, Smith [20] observes that a partial evaluator for a family of constraint logic programming languages succeeds in generating linear matchers that do not perform redundant tests, because negative information is also maintained. The same is true for Generalized Partial Computation [6], where a theorem prover is used to derive additional information from the truth or falsity of enclosing conditional tests, and for Queinnec and Ge#roy s intelligent backtracking system [19] where abstract descriptions of the matched and unmatched patterns are maintained across success and ....

Yoshihiko Futamura and Kenroku Nogi. Generalized partial computation. In Dines Bjrner, Andrei P. Ershov, and Neil D. Jones, editors, Partial Evaluation and Mixed Computation, pages 83--116. North Holland, 1988.

....language, are evaluated. In particular, the performance of the constraint based program specialiser is compared with programs resulting from other well established program transformers, including positive supercompilation [SGJ96, GS96] deforestation [Wad90] and generalized partial computation [FN88] 1.3 Overview 7 Chapter 6: Compiled Simulation by Program Specialisation Generating compiled simulations using a semantic based interpreter and a hardware design described in a hardware description language (HDL) is covered in Chapter 6. This chapter begins with an introduction to the hardware ....

....restrictions of deforestation, thus ensuring termination of positive supercompilation for arbitrary first order functional programs. Generalized Partial Computation Generalized partial computation (GPC) is an online program specialisation technique which allows full information propagation [FN88, FNT91] Environment information is represented by a set of 2.1 Program Specialisation by Partial Evaluation 16 predicates. Upon evaluation of a conditional expression, the information in the condition is added to the set of predicates and this set is passed to a theorem prover in order to decide ....

[Article contains additional citation context not shown here]

Y. Futamura and K. Nogi. Generalized Partial Computation. In D. Bjrner, A.P. Ershov, and N.D. Jones, editors, Partial Evaluation and Mixed Computation, page 133. North Holland, 1988.

....removing the general class of side e ects identi ed in the quotation above, but this currently remains a problem for future work. The presence of side e ects inhibits the application of many software engineering techniques, such as symbolic execution [7] slicing [4, 34, 13] partial evaluation [2, 16] and transformation [33, 28, 30] which typically work on side e ect free systems 2 . The presence of side e ects is also widely believed to inhibit program comprehension. For instance, Kernighan and Pike [23] advise abstinence from side e ects, in all but, well understood, special cases, ....

Futamura, Y., and Nogi, K. Generalized partial computation. In IFIP TC2 Workshop on Partial Evaluation and Mixed Computation (1987), D. Bjrner, A. P. Ershov, and N. D. Jones, Eds., North{Holland.

....program and hope it works. ffl We plan to increase the expressiveness of our meta language by adding qualified expression procedures, which Scherlis used in his thesis [18] to increase the information used to specialize programs. With these we get the power of Generalized Partial Computation [6, 21]. ....

Y. Futamura and K. Nogi. Generalized partial computation. In D. Bjorner, A. P. Ershov, and N. D. Jones, editors, Partial Evaluation and Mixed Computation, pages 133--151. North-Holland, 1988.

....to expression procedures, we get the power of qualified expression procedures without having to add any ad hoc constructs to the language. Let s say we have an assert function defined as assert p e = if p then e else error 4 With these we get the power of Generalized Partial Computation [FN88, Tak91]. The importance of this extra information for specialization of programs is discussed in Srenson et al. SGJ94] 7 (where error is equivalent to ) for which we add some syntactic sugar: p e = assert p e p e = p e e With assert and some simple laws about it, we get the power of ....

Y. Futamura and K. Nogi. Generalized partial computation. In D. Bjorner, A. P. Ershov, and N. D. Jones, editors, Partial Evaluation and Mixed Computation, pages 133--151. NorthHolland, 1988.

....specialised matcher, thereby achieving effects similar to those described in the present paper. This, however, is achieved by a non trivial rewrite of the subject program before partial evaluation is applied, thus rendering full automation difficult. In the case of Generalised Partial Computation [5], Takano has presented a transformation technique [19] that exceeds the power of both Turchin s supercompiler and perfect supercompilation. This extra power, however, stems from an unspecified theorem prover that needs to be fed the properties about primitive functions in the language, axioms for ....

Y. Futamura and K. Nogi. Generalized partial computation. In D. Bjørner, A.P. Ershov, and N.D. Jones, editors, Partial Evaluation and Mixed Computation, pages 133--151, Amsterdam, 1988. North-Holland.

....than the dynamic x. Negative knowledge may also be exploited. For instance, one knows that the dynamic x definitely does not have the value of E 1 in the false branch of the conditional above. Improvements of these kinds were of great importance in [CD89] Some systems automate such improvements [Tur86, FN88]. 52 7 HOW TO OBTAIN GOOD RESULTS WHEN USING SIMILIX 7.2.2 Dynamic choice of static values Consider an expression (car (if E 1 E 2 E 3 ) where E 1 is dynamic, but E 2 and E 3 static. The dynamic test will make the result of the conditional expression dynamic and hence no reduction of the car ....

Yoshihiko Futamura and Kenroku Nogi. Generalized partial computation. In Dines Bjørner, Andrei P. Ershov, and Neil D. Jones, editors, Partial Evaluation and Mixed Computation, pages 133--151, North-Holland, 1988.

.... loop absorption [31] partial evaluation of functionallogic languages [1] unfold fold transformation of functional programs [11] unfold fold transformation of logic programs [38] tupling [4, 29] supercompilation [39, 40] positive supercompilation [18, 35] generalized partial computation [15], deforestation [42] and online partial evaluation of functional programs [43, 33, 21] Although offline transformers (i.e. transformers making use of analyses prior to the transformation to make changes in the program ensuring termination) may fit into the description with the three phases, the ....

Y. Futamura and K. Nogi. Generalized partial computation. In Bjørner et al. [5], pages 133--151.

....term. Some extensions of partial evaluation, both of functional and logic programs, have been proposed to deal with static properties of input data, instead of static values. These extensions incorporate complex forms of reasoning based on abstract interpretation [3, 4] and theorem proving [1, 5]. By contrast, we propose a new technique which is based on the unfolding rule and a very limited form of goal replacement. On the one hand, this technique allows us to specify static properties in a precise way (unlike abstract interpretation) and, on the other hand, we avoid the use of very ....

....proposed to deal with properties of the input data. Parametrized Partial Evaluation [3] allows for the specialization of programs w.r.t. properties of the input data which are described by means of approximated representations based on abstract interpretations. Generalized Partial Computation [5] allows for the use of theorem provers for dealing with properties which restrict the domains of the functional programs to be specialized. In our approach, we avoid the use of abstract interpretations and we specify properties of the input data in a precise way. We also avoid the need for very ....

[Article contains additional citation context not shown here]

Y. Futamura and K. Nogi. Generalized partial computation. In D. Bjørner, A. P. Ershov, and N. D. Jones, editors, Partial Evaluation and Mixed Computation, pages 133--151. North-Holland, 1988.

.... = fname = Null , arity = 0, span = 3g val Leafc = fname = Leaf , arity = 1, span = 3g val Nodec = fname = Node , arity = 3, span = 3g Then the SML term Node(Leaf 9, 12, Node(Leaf 4, 7, Null) would be encoded by the following term: PCon(Nodec, PCon(Leafc, 9] 12, PCon(Nodec, PCon(Leafc,[4]) 7,PCon(Nullc, A match rule or mrule is a pair of a pattern and an (unspecified) right hand side, and a match is a sequence of match rules: type rhs match = pat rhs) list A match object is the term to be matched against a pattern. A term matches a pattern if there is a substitution ....

.... [ varc, Success 222) lamc, Success 333) appc, Success 444) Success 888) App, Switch(Sel(1, Obj) Lam, Success 555) App, Success 666) Failure) IfEq(Sel(2, Obj) Let, Success 777, IfEq(Sel(3, Obj) App, Success 999, Failure) 8 Related work Partial evaluation: Futamura and Nogi [4] showed that in principle, efficient string matchers a la Knuth Morris Pratt can be generated from a general naive matcher. This required a generalized partial evaluator which would record the outcome of previous tests and use a theorem prover to decide subsequent tests. Apparently it was not ....

Y. Futamura and K. Nogi. Generalized partial computation. In D. Bjørner, A.P. Ershov, and N.D. Jones, editors, Partial Evaluation and Mixed Computation, pages 133--151. Amsterdam: North-Holland, 1988.

....slice any program transformation may be used provided it yields a simpler program preserves the projected meaning of the original. This allows considerable flexibility. For example, conventional slicing, transformation [21] computer algebra [13] partial evaluation and mixed computation [9, 37, 38, 17] and symbolic execution [11, 12] are all candidates together with a variety of well known compiler optimisation techniques [4] such as loop unfolding, code motion, and constant propagation. The authors are currently working on the development of an algorithm for constructing syntactically ....

....are an integral part of the approach to program comprehension advocated here, as they are used as part of the algorithm for constructing syntactically unrestricted quasi static slices. The concept of a syntactically unrestricted quasi static slice is closely related to that of partial evaluation [9, 17, 38] (also known as mixed computation [16] Partial evaluation is a method of specialising a program with respect to an initial prefix of the input to a program. The process of partial evaluation produces a simplified program (called a residual program ) and a new state, oe, such that when the ....

Futamura, Y., and Nogi, K. Generalized partial computation. In IFIP TC2 Workshop on Partial Evaluation and Mixed Computation (1987), D. Bjørner, A. P. Ershov, and N. D. Jones, Eds., North--Holland.

.... set of constraints, as formalized by our inference system [14] is not only mechanizable, it is also generalizes the conditions under which partial evaluation usually takes place (cf. 3, 5, 6] In this sense it is related the notion of generalized partial computation proposed by Futamura and Nogi [4]. Constraints generalize the usual known unknown dichotomy of partial evaluation and can be implemented by symbolic values as in [20] While we have by no means fully met the challenge presented in [19] we have laid the groundwork for application of partial evaluation to the problem of component ....

Y. Futamura and K. Nogi. Generalized partial computation. In D. Bjorner, A. P. Erschov, and N. D. Jones, editors, Partial Evaluation and Mixed Computation. North--Holland, 1988.

....as well as some more dramatic optimizations. This is done by driving, i.e. unfolding and propagation of information, and generalization (Turchin, 1988) a form of abstraction which enables folding. The decision when to generalize is taken online. Generalized partial computation (GPC) due to Futamura (1988), has similar effects and power as supercompilation, but requires the use of a theorem prover. The above methodologies have been developed for functional languages. Similar methodologies are also being studied for other language paradigms, e.g. partial deduction in logic programming (Lloyd and ....

....theorem prover checks whether more than one branch is possible. If only one is possible, only that branch is taken. GPC is a powerful transformation method because it propagates predicates rather than just value information, and it assumes the (unlimited) power of a theorem prover. It was shown in (Futamura and Nogi, 1988) that this information suffices to pass the KMP test on a general matcher. Supercompilation and GPC are related, but differ in the propagation of information. While GPC propagates arbitrary predicates (logical formulas) requiring a 7 Gluck and Klimov (1993) consider perfect driving for a ....

[Article contains additional citation context not shown here]

Futamura, Y. and Nogi, K. 1988. Generalized partial computation. In Partial Evaluation and Mixed Computation, Bjørner D., Ershov A.P. and Jones N.D. (eds.), pages 133--151, North-Holland.

....enables folding. The decision when to generalize is taken online. Recent work by Gluck and Klimov has expressed the essence of driving in the context of a more traditional tail recursive language manipulating Lisp like lists [Glu93] Generalized partial computation (GPC) due to Futamura and Nogi [Fut88] and later applied to a lazy functional language [Tak91] has similar effects and power as supercompilation, but has not yet been implemented. The remainder of the paper is organized as follows. In Section 2 we introduce some terminology that allows us to discuss the quality of the output of ....

....the (unlimited) power of a theorem 8 In self application of partial evaluation one does binding time analysis of the partial evaluator; such an analysis gives better results for the environment based version because it gives better separation of static and dynamic data. prover. It was shown in [Fut88] that this information suffices to pass the KMP test on the tail recursive matcher. Supercompilation and GPC are related, but differ in the propagation of information. While the latter propagates arbitrary predicates requiring a theorem prover, supercompilation propagates structural predicates ....

[Article contains additional citation context not shown here]

Y. Futamura & K. Nogi. Generalized Partial Computation. In Partial Evaluation and Mixed Computation. Eds. A. P. Ershov, D. Bjørner & N. D. Jones, pp.133-151, North-Holland 1988.

....in the residual program; this is due to the lack of negative information in the transformation. Program transformation techniques that do propagate negative information during program specialisation include Turchin s perfect supercompilation [GK93] and Futamura s generalized partial computation [FN88]. Supercompilation passes the structural information via environments, while, in generalized partial computation, information P is propagated through arbitrary predicates, and a powerful theorem prover is used to test P or :P when necessary. These transformation techniques also pass the KMP test. ....

Y. Futamura and K. Nogi. Generalized partial computation. In A.P. Ershov D. Bjorner and N.D. Jones, editors, Partial Evaluation and Mixed Computation, page 133. North Holland, 1988.

.... copy of itself, rather than the input string; and the outcome of such a match can be decided at specialization time [28] Rather than rewriting the source program to make it keep a static track of dynamic values, one can also obtain this residual program by generalizing the partial evaluator [46, 58, 104]. For example, let us specialize the following function by letting its first parameter be 10: lambda (s d) case d [ 1) s d) 2) s d) else d] A naive strategy would yield the following residual program. lambda (d) case d [ 1) 10 d) 2) 10 d) else d] However, a ....

Y. Futamura and K. Nogi. Generalized partial computation. In Bjørner et al. [9].

....value. If it is, then by explicitly adding an assignment in the truth branch of the conditional and copying the statements which use x into both branches, the statements in the truth branch can be specialized with respect to this common value for x. This example of generalized partial computation [15, 16] has proven useful both with the Sun RPC as well as with application generation. This binding time improvement is possible because the binding time analysis is flow sensitive. The second example shows how return sensitivity is crucial to specialize the excerpt of the Sun RPC client code [21] shown ....

Y. Futamura and K. Nogi. Generalized partial computation. In D. Bjørner, A.P. Ershov, and N.D. Jones, editors, Partial Evaluation and Mixed Computation, pages 133--151. Amsterdam: North-Holland, 1988.

....(x and p) not the outcome of previously encountered dynamic tests (on n) Polyvariant specialization may be enhanced to do so, giving generalized partial evaluation. Then a theorem prover is required to decide static conditionals and to decide whether two static environments are equivalent [41]. In certain data domains and applications, less powerful methods may suffice [46] 6 Partial evaluation for other languages 6.1 Functional languages 6.1.1 First order languages Partial evaluation of a first order functional language may be done by polyvariant specialization as described in ....

Y. Futamura and K. Nogi. Generalized partial computation. In D. Bjørner, A.P. Ershov, and N.D. Jones, editors, Partial Evaluation and Mixed Computation, pages 133--151. Amsterdam: North-Holland, 1988.

No context found.

Y. Futamura, K. Nogi. Generalized partial computation. In D. Bjrner, A. P. Ershov, N. D. Jones (eds.), Partial Evaluation and Mixed Computation, 133--151. North-Holland, 1988.

....way to eliminate the ine#ciencies. Generalized Partial Computation (GPC) is a program transformation method utilizing partial information about input data, abstract data types of auxiliary functions and the logical structure of a source program. The basic idea of GPC was reported at PEMC 87 [7]. GPC uses a theorem prover to solve branching conditions including unknown variables. It also uses the prover to decide termination conditions of unfolding and conditions for correct folding. The prover uses domain information about variables, abstract data types of auxiliary functions (knowledge ....

....5.10 Pattern Matcher (matchaab) Source program matchaab(x) is a non linear pattern matcher that check if there is pattern [a, a, b] in a given text x. The residual program is a KMP type linear pattern matcher. Note that the pattern matcher used here in the source program is as naive as the one in [7, 18, 28], but more naive than the one used in [8] matchaab(x) # f ( a, a, b] x, a, a, b] x) f(p, t, p 0 , t 0 ) # if Null(p) then true else if Null(t) then false else if car(p) car(t) then f(cdr(p) cdr(t) p 0 , t 0 ) else if Null(t 0 ) then false else f(p 0 , cdr(t 0 ) p 0 , cdr(t 0 ....

Futamura, Y., Nogi, K.: Generalized partial computation. in Bjrner, D., Ershov, A.P., Jones, N.D. (eds.): Partial Evaluation and Mixed Computation. NorthHolland, Amsterdam (1988) 133--151.

No context found.

FUTAMURA, Y., AND NOGI, K. Generalized partial computation. In IFIP TC2 Workshop on Partial Evaluation and Mixed Computation (1987), D. Bjrner, A. P. Ershov, and N. D. Jones, Eds., North--Holland, pp. 133--151.

No context found.

Y. Futamura and K. Nogi. Generalized Partial Computation. In D. Bjrner, A.P. Ershov, and N.D. Jones, editors, Partial Evaluation and Mixed Computation, page 133. North Holland, 1988.

No context found.

Yoshihiko Futamura and Kenroku Nogi. Generalized partial computation. In Dines Bjrner, Andrei P. Ershov, and Neil D. Jones, editors, Partial Evaluation and Mixed Computation, pages 133--151. North-Holland, 1988.

No context found.

Yoshihiko Futamura and Kenroku Nogi. Generalized partial computation. In Bjrner et al. [9], pages 133-151.

No context found.

Yoshihiko Futamura and Kenroku Nogi. Generalized partial computation. In Dines Bjrner, Andrei P. Ershov, and Neil D. Jones, editors, Partial Evaluation and Mixed Computation, pages 133--151. NorthHolland, 1988.

No context found.

Futamura, Y., Nogi, K.: Generalized partial computation. In Bjrner, D., Ershov, A.P., Jones, N.D., eds.: Proceedings of the IFIP Workshop on Partial Evaluation and Mixed Computation, North-Holland (1987)

No context found.

Y. Futamura and K. Nogi. Generalized partial computation. In D. Bjrner, A. P. Ershov, and N. D. Jones, editors, Proceedings of the IFIP Workshop on Partial Evaluation and Mixed Computation. North-Holland, 1987.

No context found.

) Futamura, Y. and Nogi, K., "Generalized Partial Computation," in Partial Evaluation and Mixed Computation (Bj0rner, D., Ershov, A. P. and Jones, N. D. eds.), North-Holland, Amsterdam, pp. 133-151, 1988.

No context found.

Y. Futamura, K. Nogi, Generalized Partial Computation, in: [17], pp. 133-152

No context found.

Y. Futamura and K. Nogi. Generalized partial computation. In D. Bjørner, A. Ershov, and N. Jones, editors, Partial Evaluation and Mixed Computation, pages 133--151. Amsterdam: North-Holland, 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