Stefano crespi reghizzi | Politecnico di Milano (original) (raw)
Papers by Stefano crespi reghizzi
Journal of the ACM, 1978
... 573 It is immediately observable that if we parenthesize a t'mite state grammar which ge... more ... 573 It is immediately observable that if we parenthesize a t'mite state grammar which generates a language that is noncounting according to [15], the parenthesis language thus obtained is noncounting according to our definition, and vice versa. ...
Electronic Proceedings in Theoretical Computer Science
Floyd's Operator Precedence (OP) languages are a deterministic context-free family having many de... more Floyd's Operator Precedence (OP) languages are a deterministic context-free family having many desirable properties. They are locally and parallely parsable, and languages having a compatible structure are closed under Boolean operations, concatenation and star; they properly include the family of Visibly Pushdown (or Input Driven) languages. OP languages are based on three relations between any two consecutive terminal symbols, which assign syntax structure to words. We extend such relations to k-tuples of consecutive terminal symbols, by using the model of strictly locally testable regular languages of order k ≥ 3. The new corresponding class of Higher-order Operator Precedence languages (HOP) properly includes the OP languages, and it is still included in the deterministic (also in reverse) context free family. We prove Boolean closure for each subfamily of structurally compatible HOP languages. In each subfamily, the top language is called max-language. We show that such languages are defined by a simple cancellation rule and we prove several properties, in particular that max-languages make an infinite hierarchy ordered by parameter k. HOP languages are a candidate for replacing OP languages in the various applications where they have have been successful though sometimes too restrictive.
ABSTRACT Associative Language Descriptions are a recent grammar model, theoretically less powerfu... more ABSTRACT Associative Language Descriptions are a recent grammar model, theoretically less powerful than Context Free grammars, but adequate for describing the syntax of programming languages. ALD do not use nonterminal symbols, but rely on permissible contexts for specifying valid syntax trees. In order to assess ALD adequacy, we analyze the descriptional complexity of structurally equivalent CF and ALD grammars, finding comparable measures. The compression obtained using CF copy rules is matched by context inheritance in ALD. The family of hierarchical parentheses languages, an abstract paradigm of HTML, and of expressions with operator precedences are studied in detail. A complete ALD grammar of Pascal is presented to testify of the practicality of the ALD approach. Keywords: Context Free Grammars, Associative Grammars, Grammar Size, Context Inheritance, Descriptional Complexity. 1.
ABSTRACT A generalization of context-free grammars and languages is presented. Generalized contex... more ABSTRACT A generalization of context-free grammars and languages is presented. Generalized context-free, or gcf, languages strictly include the cf ones but, unlike most past proposed extensions, remain semilinear. Gcf grammars combine the usage of the depthfirst rewriting discipline with the breadth-first one; the recognizers of gcf languages have tapes handled as pushdown (LIFO) or queue (FIFO) storages. This paper proves for gcf languages a partial analogous of the classical Chomsky-Schutzenberger theorem of the cf ones: every gcf language is the hom. image of the intersection of the generalized Dyck language with a regular language. The proof requires the computation of the adjacencies of letters in the strings of gcf languages. This problem is solved by means of algebraic methods, based on an adjacency-preserving compression of strings. 1 Introduction Many variations on the theme of formal grammars have been advanced since Chomsky 's classification of grammar types [20], but the context-fr...
This paper has the goals (1) of unifying top-down parsing with shift-reduce parsing to yield a si... more This paper has the goals (1) of unifying top-down parsing with shift-reduce parsing to yield a single simple and consistent framework, and (2) of producing provably correct parsing methods, deterministic as well as tabular ones, for extended context-free grammars (EBNF) represented as state-transition networks. Departing from the traditional way of presenting as independent algorithms the deterministic bottom-up LR(1), the top-down LL(1) and the general tabular (Earley) parsers, we unify them in a coherent minimalist framework. We present a simple general construction method for EBNF ELR(1) parsers, where the new category of convergence conflicts is added to the classical shift-reduce and reduce-reduce conflicts; we prove its correctness and show two implementations by deterministic push-down machines and by vector-stack machines, the latter to be also used for Earley parsers. Then the Beatty's theoretical characterization of LL(1) grammars is adapted to derive the extended ELL(1 parsing method, first by minimizing the ELR(1) parser and then by simplifying its state information. Through using the same notations in the ELR(1) case, the extended Earley parser is obtained. Since all the parsers operate on compatible representations, it is feasible to combine them into mixed mode algorithms.
Theoretical Computer Science, 2015
We establish the relation between two language recognition models that use counters and operate i... more We establish the relation between two language recognition models that use counters and operate in real-time: Greibach's partially blind machines operating in real time (RT-PBLIND), which recognize Petri Net languages, and the consensually regular (CREG) language model of the authors. The latter is based on synchronized computational threads of a finite automaton, where at each step one thread acts as the leader and all other threads as followers. We introduce two new normal forms of RT-PBLIND machines (and Petri Nets), such that counter operations are scheduled and rarefied, and transitions are quasi-deterministic, i.e., the finite automaton obtained by eliminating counter moves is deterministic. We prove that the CREG family can simulate any normalized RT-PBLIND machine, but it also contains the non-RT-PBLIND language {a n b n | n > 1} * .
ABSTRACT . This original method for specifying and checking the sequences of events taking place ... more ABSTRACT . This original method for specifying and checking the sequences of events taking place in process scheduling brings the classical syntax-directed approach of compilation to this new area. The formal language of scheduling events cannot be specified by BNF grammars, but we use instead the Augmented BNF grammars, which combine breadth-- first and depth-first derivations. Their recognizers feature one or more FIFO or LIFO tapes. The basic scheduling policies are covered: FCFS, time-slicing, mutex. Combined policies, such as readers/writers and background /foreground, are obtainable by composition. Constraints on the minimum number of data structures (i.e. queues) for priority scheduling policies may be proved by using a pumping lemma. The construction of schedule checkers is presented in the form of augmented LL(1) parsers. For scheduling algorithms, such as shortest job first, which depend on parameters and in particular on time, a syntax-directed approach is proposed, whi...
Milan J Math, 1981
... fissato, i particolari valori il/2 .../~ influiseono sulla velocita dell'algoritmo, cio~... more ... fissato, i particolari valori il/2 .../~ influiseono sulla velocita dell'algoritmo, cio~ per ogni caso I, di//il tempo di ealeolo sara in generale funzione di I,, oltre che din. ... Si prefe-risce perci6 limitarsi a due easi significativi, noti come no~na del caso peggiore e norma del caso medio. ...
Lecture Notes in Computer Science, 1988
Without Abstract
In this work we study some properties of integer compositions in connection with the recognition ... more In this work we study some properties of integer compositions in connection with the recognition of rational trace languages. In particular, we introduce some operations defined on integer compositions and present procedures for their computation that work in linear or in quadratic time. These procedures turn out to be useful in the analysis of syntactic trees of certain regular expressions, called repeat-until expressions, which intuitively represent programs of instructions nested in repeat-until loops. Our main aim is to show how, in some cases, such an analysis allows us to design algorithms for the recognition of (rational) trace languages defined by repeat-until expressions, which work in quadratic time independently of the concurrency relation.
Proceedings of the 2006 ACM symposium on Applied computing - SAC '06, 2006
... 7. REFERENCES [1] Giovanni Agosta, Stefano Crespi-Reghizzi, Gerlando Falauto, and Martino Syk... more ... 7. REFERENCES [1] Giovanni Agosta, Stefano Crespi-Reghizzi, Gerlando Falauto, and Martino Sykora. JIST: Just-in-Time Scheduling Translation for Parallel Processors. In ISPDC/HeteroPar, pages 122{132. IEEE Computer Society, 2004. [2] John Aycock. ...
ACM SIGPLAN Notices, 1983
A methodology for designing and implementing program construction systems using relational data b... more A methodology for designing and implementing program construction systems using relational data bases is presented. Relations are the only data structures used by them and in the definition of interfaces between subsystems. A special-purpose relational database manager (RDB) has been designed for this application.Two large projects using this approach are described. ART (Ada-Relational-Translator) is an experimental compiler-interpreter for ADA in which all subsystems, including parser, semantic analyzer, interpreter and kernel, use relations as their only data structure; the relational approach has been pushed to the upmost to achieve fast prototyping. MML (Multi-Micro Line) is a tool-set for constructing program programs for multi-microprocessors targets, in which relations are used for allocation and configuration control. Both experiences confirm the validity of the approach for managing team work in evolving projects.
Proceedings of third international workshop on MapReduce and its Applications Date - MapReduce '12, 2012
ABSTRACT Research has proved that machine learning and iterative compilation techniques can be pr... more ABSTRACT Research has proved that machine learning and iterative compilation techniques can be profitable when applied to compilers to improve the optimizations they perform on programs. Unfortunately, these techniques are hampered by the long training times they require. This paper shows that parallel execution of multiple training runs can be naturally mapped on the MapReduce programming model and is effective in reducing execution times for iterative compilation. The presented technique allows parallel execution on multiple identical worker nodes or on a single machine by splitting its resources. Experimental results show that an almost-linear speedup can be obtained.
Texts in Computer Science, 2009
Lecture Notes in Computer Science, 2015
and took place at Umeå Folkets hus during August 18-21, 2015. The CIAA conference series is the m... more and took place at Umeå Folkets hus during August 18-21, 2015. The CIAA conference series is the major international venue for the dissemination of new results in the implementation, application, and theory of automata. The previous 19 conferences were held in various locations all around the globe:
The classical logical antinomy known as Richard-Berry paradox is combined with plausible assumpti... more The classical logical antinomy known as Richard-Berry paradox is combined with plausible assumptions about the size i.e. the descriptional complexity of Turing machines formalizing certain sentences, to show that formalization of language leads to contradiction.
Texts in Computer Science, 2009
Journal of the ACM, 1978
... 573 It is immediately observable that if we parenthesize a t'mite state grammar which ge... more ... 573 It is immediately observable that if we parenthesize a t'mite state grammar which generates a language that is noncounting according to [15], the parenthesis language thus obtained is noncounting according to our definition, and vice versa. ...
Electronic Proceedings in Theoretical Computer Science
Floyd's Operator Precedence (OP) languages are a deterministic context-free family having many de... more Floyd's Operator Precedence (OP) languages are a deterministic context-free family having many desirable properties. They are locally and parallely parsable, and languages having a compatible structure are closed under Boolean operations, concatenation and star; they properly include the family of Visibly Pushdown (or Input Driven) languages. OP languages are based on three relations between any two consecutive terminal symbols, which assign syntax structure to words. We extend such relations to k-tuples of consecutive terminal symbols, by using the model of strictly locally testable regular languages of order k ≥ 3. The new corresponding class of Higher-order Operator Precedence languages (HOP) properly includes the OP languages, and it is still included in the deterministic (also in reverse) context free family. We prove Boolean closure for each subfamily of structurally compatible HOP languages. In each subfamily, the top language is called max-language. We show that such languages are defined by a simple cancellation rule and we prove several properties, in particular that max-languages make an infinite hierarchy ordered by parameter k. HOP languages are a candidate for replacing OP languages in the various applications where they have have been successful though sometimes too restrictive.
ABSTRACT Associative Language Descriptions are a recent grammar model, theoretically less powerfu... more ABSTRACT Associative Language Descriptions are a recent grammar model, theoretically less powerful than Context Free grammars, but adequate for describing the syntax of programming languages. ALD do not use nonterminal symbols, but rely on permissible contexts for specifying valid syntax trees. In order to assess ALD adequacy, we analyze the descriptional complexity of structurally equivalent CF and ALD grammars, finding comparable measures. The compression obtained using CF copy rules is matched by context inheritance in ALD. The family of hierarchical parentheses languages, an abstract paradigm of HTML, and of expressions with operator precedences are studied in detail. A complete ALD grammar of Pascal is presented to testify of the practicality of the ALD approach. Keywords: Context Free Grammars, Associative Grammars, Grammar Size, Context Inheritance, Descriptional Complexity. 1.
ABSTRACT A generalization of context-free grammars and languages is presented. Generalized contex... more ABSTRACT A generalization of context-free grammars and languages is presented. Generalized context-free, or gcf, languages strictly include the cf ones but, unlike most past proposed extensions, remain semilinear. Gcf grammars combine the usage of the depthfirst rewriting discipline with the breadth-first one; the recognizers of gcf languages have tapes handled as pushdown (LIFO) or queue (FIFO) storages. This paper proves for gcf languages a partial analogous of the classical Chomsky-Schutzenberger theorem of the cf ones: every gcf language is the hom. image of the intersection of the generalized Dyck language with a regular language. The proof requires the computation of the adjacencies of letters in the strings of gcf languages. This problem is solved by means of algebraic methods, based on an adjacency-preserving compression of strings. 1 Introduction Many variations on the theme of formal grammars have been advanced since Chomsky 's classification of grammar types [20], but the context-fr...
This paper has the goals (1) of unifying top-down parsing with shift-reduce parsing to yield a si... more This paper has the goals (1) of unifying top-down parsing with shift-reduce parsing to yield a single simple and consistent framework, and (2) of producing provably correct parsing methods, deterministic as well as tabular ones, for extended context-free grammars (EBNF) represented as state-transition networks. Departing from the traditional way of presenting as independent algorithms the deterministic bottom-up LR(1), the top-down LL(1) and the general tabular (Earley) parsers, we unify them in a coherent minimalist framework. We present a simple general construction method for EBNF ELR(1) parsers, where the new category of convergence conflicts is added to the classical shift-reduce and reduce-reduce conflicts; we prove its correctness and show two implementations by deterministic push-down machines and by vector-stack machines, the latter to be also used for Earley parsers. Then the Beatty's theoretical characterization of LL(1) grammars is adapted to derive the extended ELL(1 parsing method, first by minimizing the ELR(1) parser and then by simplifying its state information. Through using the same notations in the ELR(1) case, the extended Earley parser is obtained. Since all the parsers operate on compatible representations, it is feasible to combine them into mixed mode algorithms.
Theoretical Computer Science, 2015
We establish the relation between two language recognition models that use counters and operate i... more We establish the relation between two language recognition models that use counters and operate in real-time: Greibach's partially blind machines operating in real time (RT-PBLIND), which recognize Petri Net languages, and the consensually regular (CREG) language model of the authors. The latter is based on synchronized computational threads of a finite automaton, where at each step one thread acts as the leader and all other threads as followers. We introduce two new normal forms of RT-PBLIND machines (and Petri Nets), such that counter operations are scheduled and rarefied, and transitions are quasi-deterministic, i.e., the finite automaton obtained by eliminating counter moves is deterministic. We prove that the CREG family can simulate any normalized RT-PBLIND machine, but it also contains the non-RT-PBLIND language {a n b n | n > 1} * .
ABSTRACT . This original method for specifying and checking the sequences of events taking place ... more ABSTRACT . This original method for specifying and checking the sequences of events taking place in process scheduling brings the classical syntax-directed approach of compilation to this new area. The formal language of scheduling events cannot be specified by BNF grammars, but we use instead the Augmented BNF grammars, which combine breadth-- first and depth-first derivations. Their recognizers feature one or more FIFO or LIFO tapes. The basic scheduling policies are covered: FCFS, time-slicing, mutex. Combined policies, such as readers/writers and background /foreground, are obtainable by composition. Constraints on the minimum number of data structures (i.e. queues) for priority scheduling policies may be proved by using a pumping lemma. The construction of schedule checkers is presented in the form of augmented LL(1) parsers. For scheduling algorithms, such as shortest job first, which depend on parameters and in particular on time, a syntax-directed approach is proposed, whi...
Milan J Math, 1981
... fissato, i particolari valori il/2 .../~ influiseono sulla velocita dell'algoritmo, cio~... more ... fissato, i particolari valori il/2 .../~ influiseono sulla velocita dell'algoritmo, cio~ per ogni caso I, di//il tempo di ealeolo sara in generale funzione di I,, oltre che din. ... Si prefe-risce perci6 limitarsi a due easi significativi, noti come no~na del caso peggiore e norma del caso medio. ...
Lecture Notes in Computer Science, 1988
Without Abstract
In this work we study some properties of integer compositions in connection with the recognition ... more In this work we study some properties of integer compositions in connection with the recognition of rational trace languages. In particular, we introduce some operations defined on integer compositions and present procedures for their computation that work in linear or in quadratic time. These procedures turn out to be useful in the analysis of syntactic trees of certain regular expressions, called repeat-until expressions, which intuitively represent programs of instructions nested in repeat-until loops. Our main aim is to show how, in some cases, such an analysis allows us to design algorithms for the recognition of (rational) trace languages defined by repeat-until expressions, which work in quadratic time independently of the concurrency relation.
Proceedings of the 2006 ACM symposium on Applied computing - SAC '06, 2006
... 7. REFERENCES [1] Giovanni Agosta, Stefano Crespi-Reghizzi, Gerlando Falauto, and Martino Syk... more ... 7. REFERENCES [1] Giovanni Agosta, Stefano Crespi-Reghizzi, Gerlando Falauto, and Martino Sykora. JIST: Just-in-Time Scheduling Translation for Parallel Processors. In ISPDC/HeteroPar, pages 122{132. IEEE Computer Society, 2004. [2] John Aycock. ...
ACM SIGPLAN Notices, 1983
A methodology for designing and implementing program construction systems using relational data b... more A methodology for designing and implementing program construction systems using relational data bases is presented. Relations are the only data structures used by them and in the definition of interfaces between subsystems. A special-purpose relational database manager (RDB) has been designed for this application.Two large projects using this approach are described. ART (Ada-Relational-Translator) is an experimental compiler-interpreter for ADA in which all subsystems, including parser, semantic analyzer, interpreter and kernel, use relations as their only data structure; the relational approach has been pushed to the upmost to achieve fast prototyping. MML (Multi-Micro Line) is a tool-set for constructing program programs for multi-microprocessors targets, in which relations are used for allocation and configuration control. Both experiences confirm the validity of the approach for managing team work in evolving projects.
Proceedings of third international workshop on MapReduce and its Applications Date - MapReduce '12, 2012
ABSTRACT Research has proved that machine learning and iterative compilation techniques can be pr... more ABSTRACT Research has proved that machine learning and iterative compilation techniques can be profitable when applied to compilers to improve the optimizations they perform on programs. Unfortunately, these techniques are hampered by the long training times they require. This paper shows that parallel execution of multiple training runs can be naturally mapped on the MapReduce programming model and is effective in reducing execution times for iterative compilation. The presented technique allows parallel execution on multiple identical worker nodes or on a single machine by splitting its resources. Experimental results show that an almost-linear speedup can be obtained.
Texts in Computer Science, 2009
Lecture Notes in Computer Science, 2015
and took place at Umeå Folkets hus during August 18-21, 2015. The CIAA conference series is the m... more and took place at Umeå Folkets hus during August 18-21, 2015. The CIAA conference series is the major international venue for the dissemination of new results in the implementation, application, and theory of automata. The previous 19 conferences were held in various locations all around the globe:
The classical logical antinomy known as Richard-Berry paradox is combined with plausible assumpti... more The classical logical antinomy known as Richard-Berry paradox is combined with plausible assumptions about the size i.e. the descriptional complexity of Turing machines formalizing certain sentences, to show that formalization of language leads to contradiction.
Texts in Computer Science, 2009