## Program specialisation and abstract interpretation reconciled (1998)

### Cached

### Download Links

- [www.stups.uni-duesseldorf.de]
- [www.stups.uni-duesseldorf.de]
- DBLP

### Other Repositories/Bibliography

Venue: | In Joint International Conference and Symposium on Logic Programming |

Citations: | 27 - 13 self |

### BibTeX

@INPROCEEDINGS{Leuschel98programspecialisation,

author = {Michael Leuschel},

title = {Program specialisation and abstract interpretation reconciled},

booktitle = {In Joint International Conference and Symposium on Logic Programming},

year = {1998},

pages = {220--234},

publisher = {MIT Press}

}

### OpenURL

### Abstract

We clarify the relationship between abstract interpretation and program specialisation in the context of logic programming. We present a generic top-down abstract specialisation framework, along with a generic correctness result, into which a lot of the existing specialisation techniques can be cast. The framework also shows how these techniques can be further improved by moving to more refined abstract domains. It, however, also highlights inherent limitations shared by all these approaches. In order to overcome them, and to fully unify program specialisation with abstract interpretation, we also develop a generic combined bottom-up/top-down framework, which allows specialisation and analysis outside the reach of existing techniques. 1

### Citations

1902 |
Foundations of Logic Programming
- Lloyd
- 1987
(Show Context)
Citation Context ..., but which we then fully integrate with Algorithm 2.10 in Section 4. In a bottom-up setting we need, instead of an abstraction of the unfolding operation, an abstraction of the bottom-up TP operator =-=[1, 19]-=-, or better its non-ground version (to capture the C-semantics and thus the computed answers). The (non-ground) TP operator maps interpretations to interpretations, where an interpretation is usually ... |

431 |
Logic programing, in
- Apt
- 1990
(Show Context)
Citation Context ..., but which we then fully integrate with Algorithm 2.10 in Section 4. In a bottom-up setting we need, instead of an abstraction of the unfolding operation, an abstraction of the bottom-up TP operator =-=[1, 19]-=-, or better its non-ground version (to capture the C-semantics and thus the computed answers). The (non-ground) TP operator maps interpretations to interpretations, where an interpretation is usually ... |

295 | Abstract interpretation and application to logic programs
- Cousot, Cousot
- 1992
(Show Context)
Citation Context ...p a generic combined bottom-up/top-down framework, which allows specialisation and analysis outside the reach of existing techniques. 1 Introduction At first sight abstract interpretation (see, e.g., =-=[5, 3]-=-) and program specialisation (see, e.g., [7]) might appear to be completely unrelated techniques: abstract interpretation focusses on correct and precise analysis, while the main goal of program speci... |

242 |
Partial Evaluation in Logic Programming
- Lloyd, Shepherdson
- 1991
(Show Context)
Citation Context ...ee, e.g., [16, 11, 22]. Indeed, for good specialisation to take place, program specialisers have to perform some form of analysis. For instance, the incomplete SLD-trees produced by partial deduction =-=[20, 7]-=- can be seen as complete (given the closedness condition of [20]) description of the top-down computation-flow. In this paper we want to substantiate this intuition and make the link to abstract inter... |

196 |
A practical framework for the abstract interpretation of logic programs
- Bruynooghe
- 1991
(Show Context)
Citation Context ...p a generic combined bottom-up/top-down framework, which allows specialisation and analysis outside the reach of existing techniques. 1 Introduction At first sight abstract interpretation (see, e.g., =-=[5, 3]-=-) and program specialisation (see, e.g., [7]) might appear to be completely unrelated techniques: abstract interpretation focusses on correct and precise analysis, while the main goal of program speci... |

151 | Tutorial on Specialisation of Logic Programs
- Gallagher
- 1993
(Show Context)
Citation Context ...rk, which allows specialisation and analysis outside the reach of existing techniques. 1 Introduction At first sight abstract interpretation (see, e.g., [5, 3]) and program specialisation (see, e.g., =-=[7]-=-) might appear to be completely unrelated techniques: abstract interpretation focusses on correct and precise analysis, while the main goal of program specialisation is to produce more efficient resid... |

59 |
Controlling generalisation and polyvariance in partial deduction of normal logic programs
- Leuschel, Martens, et al.
- 1998
(Show Context)
Citation Context .... We therefore present a generic (augmented) top-down abstract interpretation framework in which most of the specialisation techniques (such as partial deduction [20, 7], ecological partial deduction =-=[18, 13]-=-, constrained partial deduction [15], conjunctive partial deduction [17, 10]) can be cast. It also paves the way for more refined and 1 The work was done while the author was Post-doctoral Fellow of t... |

56 |
Query Evaluation in Recursive Databases: Bottom-Up and Top-Down Heconciled, Data Knowledge Eng
- Bry
- 1990
(Show Context)
Citation Context ...e can actually also express techniques based upon tabling (OLDT [12] or even EOLDT [2]) in a slight extension of our framework. In other words, the reconciliation of bottom-up and top-down evaluation =-=[4]-=- is then just a special case of our reconciliation of specialisation and analysis. In [14] the reader can find how the reverse-last open problem from [16, 13] can be solved in our framework in a rathe... |

38 |
Most specific logic programs
- Marriott, Naish, et al.
- 1988
(Show Context)
Citation Context ...eady identified in, e.g., [16, 13]): the lack of side-ways information passing and of inference of global success information. Recent techniques [16, 23, 22] (as well as some earlier attempts such as =-=[21]-=- and [9, 6]) have tried to overcome these limitations by incorporating bottom-up abstract interpretation techniques. However, we feel that a fully satisfactory integration of program specialisation wi... |

35 | Schreye. Logic program specialisation: How to be more specific
- Leuschel, De
- 1996
(Show Context)
Citation Context ...lt that there is a close relationship between abstract interpretation and program specialisation and, recently, there has been a lot of interest in the integration of these two techniques (see, e.g., =-=[16, 11, 22]-=-. Indeed, for good specialisation to take place, program specialisers have to perform some form of analysis. For instance, the incomplete SLD-trees produced by partial deduction [20, 7] can be seen as... |

33 | Controlling Conjunctive Partial Deduction of Definite Logic Programs
- Glück, Jørgensen, et al.
- 1996
(Show Context)
Citation Context ...n framework in which most of the specialisation techniques (such as partial deduction [20, 7], ecological partial deduction [18, 13], constrained partial deduction [15], conjunctive partial deduction =-=[17, 10]-=-) can be cast. It also paves the way for more refined and 1 The work was done while the author was Post-doctoral Fellow of the Belgian Fund for Scientific Research (F.W.O.-Vlaanderen). and visiting DI... |

32 |
The applicability of logic program analysis and transformation to theorem proving
- Waal, Gallagher
- 1994
(Show Context)
Citation Context ...tified in, e.g., [16, 13]): the lack of side-ways information passing and of inference of global success information. Recent techniques [16, 23, 22] (as well as some earlier attempts such as [21] and =-=[9, 6]-=-) have tried to overcome these limitations by incorporating bottom-up abstract interpretation techniques. However, we feel that a fully satisfactory integration of program specialisation with abstract... |

32 | Waal. A conceptual embedding of folding into partial deduction: Towards a maximal integration - Leuschel, Schreye, et al. - 1996 |

26 |
Metacomputation: Metasystem transitions plus supercompilation
- Turchin
- 1996
(Show Context)
Citation Context ...hnique we are aware of, was able to solve this problem. It is also possible to use the same approach to prove inductive theorems in a much less ad-hoc (and more generally reusable manner) than, e.g., =-=[24]-=-. We also believe that automation of this approach is feasible and endeavour to incorporate these possibilities into the ecce partial deduction system in the not too distant future. We believe that th... |

25 | Advanced Techniques for Logic Program Specialisation
- Leuschel
- 1997
(Show Context)
Citation Context .... We therefore present a generic (augmented) top-down abstract interpretation framework in which most of the specialisation techniques (such as partial deduction [20, 7], ecological partial deduction =-=[18, 13]-=-, constrained partial deduction [15], conjunctive partial deduction [17, 10]) can be cast. It also paves the way for more refined and 1 The work was done while the author was Post-doctoral Fellow of t... |

22 |
Combining abstract interpretation and partial evaluation
- Jones
- 1997
(Show Context)
Citation Context ...lt that there is a close relationship between abstract interpretation and program specialisation and, recently, there has been a lot of interest in the integration of these two techniques (see, e.g., =-=[16, 11, 22]-=-. Indeed, for good specialisation to take place, program specialisers have to perform some form of analysis. For instance, the incomplete SLD-trees produced by partial deduction [20, 7] can be seen as... |

21 | Schreye. Constrained partial deduction and the preservation of characteristic trees
- Leuschel, De
- 1998
(Show Context)
Citation Context ...ted) top-down abstract interpretation framework in which most of the specialisation techniques (such as partial deduction [20, 7], ecological partial deduction [18, 13], constrained partial deduction =-=[15]-=-, conjunctive partial deduction [17, 10]) can be cast. It also paves the way for more refined and 1 The work was done while the author was Post-doctoral Fellow of the Belgian Fund for Scientific Resea... |

18 |
Some Low Level Transformations of Logic Pro- grams
- Gallagher, Bruynooghe
- 1990
(Show Context)
Citation Context ...ally msg(γ(A)), instead of the skeleton ⌈A⌉ for the definition of ρA. This, however, makes the exposition more tricky and would detract from the main points of the paper. Anyway, one can always apply =-=[8]-=- (as well as RAF, see, e.g., [13]) as a post-processing. 2.5 A Generic Correctness Result and Algorithm We can now present a very general correctness result. Theorem 2.9 Let P ′ be an abstract partial... |

16 |
de Waal. Deletion of redundant unary type predicates from logic programs
- Gallagher, A
- 1992
(Show Context)
Citation Context ...tified in, e.g., [16, 13]): the lack of side-ways information passing and of inference of global success information. Recent techniques [16, 23, 22] (as well as some earlier attempts such as [21] and =-=[9, 6]-=-) have tried to overcome these limitations by incorporating bottom-up abstract interpretation techniques. However, we feel that a fully satisfactory integration of program specialisation with abstract... |

13 | Towards Integrating Partial Evaluation in a Specialization Framework based on Generic Abstract Interpretation
- Puebla, Gallagher, et al.
- 1997
(Show Context)
Citation Context ...lt that there is a close relationship between abstract interpretation and program specialisation and, recently, there has been a lot of interest in the integration of these two techniques (see, e.g., =-=[16, 11, 22]-=-. Indeed, for good specialisation to take place, program specialisers have to perform some form of analysis. For instance, the incomplete SLD-trees produced by partial deduction [20, 7] can be seen as... |

13 | Abstract Specialization and its Application to Program Parallelization - Puebla, Hermenegildo - 1997 |

7 |
OLDT-based abstract interpretation
- Kanamori, Kawamura
- 1993
(Show Context)
Citation Context ... to instantiate them) or [22] (which cannot perform deforestation or tupling as it is restricted to specialising atoms individually). One can actually also express techniques based upon tabling (OLDT =-=[12]-=- or even EOLDT [2]) in a slight extension of our framework. In other words, the reconciliation of bottom-up and top-down evaluation [4] is then just a special case of our reconciliation of specialisat... |

6 | Deriving Fold~Unfold Transformations of Logic Programs Using Eztended OLDT-based Abstract
- Boulanger, Bruynooghe
- 1993
(Show Context)
Citation Context ...m) or [22] (which cannot perform deforestation or tupling as it is restricted to specialising atoms individually). One can actually also express techniques based upon tabling (OLDT [12] or even EOLDT =-=[2]-=-) in a slight extension of our framework. In other words, the reconciliation of bottom-up and top-down evaluation [4] is then just a special case of our reconciliation of specialisation and analysis. ... |