Rewrite Step Research Articles

Optimizing term rewriting with creeper trace transducers

In the context of functional programming/term normalization algorithms we discuss the optimization problem of constructing the result of a sequence of rewrite steps, without computing all the intermediate terms. From a rewrite system we construct a creeper trace transducer, which reads a sequence of backwards overlapping rewrite steps while producing the desired answer. The transducer writes each symbol of the output only once, skipping overlap between each pair of subsequent rules. In some cases a part of the trace can be disregarded altogether.

Optical reading of multistate nonvolatile oxide memories based on the switchable ferroelectric photovoltaic effect

Intensive research into functional oxides has been triggered by the quest for a solid-state universal memory with high-storage density, non-volatility, high read/write speed, and random access. The ferroelectric random-access memory (FeRAM), in which the information is stored in the spontaneous ferroelectric polarization of the material, offers great promise as nonvolatile and multistate memory, but its destructive electrical reading step requires a rewrite step after each reading, increasing energy consumption. As an alternative, optical nondestructive readout is based on the ferroelectric polarization dependence of the photovoltaic response in materials and has been reported in two-states ferroelectric memories and multistate devices with limited photocurrent switchability due to asymmetric interfacial effects. In this work, we report a nonvolatile oxide memory device based on a symmetric heterostructure with eight stable and well-controlled remanent polarization (Pr) states, written electrically by voltage pulse and read optically through polarization-dependent short-circuit photocurrent Isc or open circuit photovoltage Voc. This symmetric capacitor demonstrates a clear proportionality between Isc (Voc) and Pr, allowing to achieve a 100% switchability of the photovoltaic response. The memory devices based on 3-bit data storage show good performance in terms of data retention, fatigue behavior, and repeatability of writing and reading cycles. Thanks to the very high sensitivity of the optical reading method, the number of states could largely exceed eight, being limited only by the electrical writing step precision. These results are particularly exciting for the development of next-generation ferroelectric memory devices with increased memory storage density and lower power consumption.

Languages Accepted by Weighted Restarting Automata*

Weighted restarting automata have been introduced to study quantitative aspects of computations of restarting automata. In earlier works we studied the classes of functions and relations that are computed by weighted restarting automata. Here we use them to define classes of formal languages by restricting the weight associated to a given input word through an additional requirement. In this way, weighted restarting automata can be used as language acceptors. First, we show that by using the notion of acceptance relative to the tropical semiring, we can avoid the use of auxiliary symbols. Furthermore, a certain type of word-weighted restarting automata turns out to be equivalent to non-forgetting restarting automata, and another class of languages accepted by word-weighted restarting automata is shown to be closed under the operation of intersection. This is the first result that shows that a class of languages defined in terms of a quite general class of restarting automata is closed under intersection. Finally, we prove that the restarting automata that are allowed to use auxiliary symbols in a rewrite step, and to keep on reading after performing a rewrite step can be simulated by regular-weighted restarting automata that cannot do this.

Algebraic graph rewriting with controlled embedding

Graph transformation is a specification technique suitable for a wide range of applications, specially the ones that require a sophisticated notion of state. In graph transformation, states are represented by graphs and actions are specified by rules. Most algebraic approaches to graph transformation proposed in the literature ensure that if an item is preserved by a rule, so are its connections with the graph where it is embedded. But there are applications in which it is desirable to specify different embeddings. For example when cloning an item, there may be a need to handle the original and the copy in different ways. We propose a new algebraic approach to graph transformation, AGREE: Algebraic Graph Rewriting with controllEd Embedding, where rules allow one to specify how the embedding should be carried out. We define this approach in the framework of classified categories which are categories endowed with partial map classifiers. This new approach leads to graph transformations in which effects may be non-local, e.g. a rewrite step may alter a node of the host graph which is outside the image of the left-hand side of the considered rule. We propose a syntactic condition on AGREE rules which guarantees the locality of transformations. We also compare AGREE with other algebraic approaches to graph transformation.

SPIRAL: Extreme Performance Portability

In this paper, we address the question of how to automatically map computational kernels to highly efficient code for a wide range of computing platforms and establish the correctness of the synthesized code. More specifically, we focus on two fundamental problems that software developers are faced with: performance portability across the ever-changing landscape of parallel platforms and correctness guarantees for sophisticated floating-point code. The problem is approached as follows: We develop a formal framework to capture computational algorithms, computing platforms, and program transformations of interest, using a unifying mathematical formalism we call operator language (OL). Then we cast the problem of synthesizing highly optimized computational kernels for a given machine as a strongly constrained optimization problem that is solved by search and a multistage rewriting system. Since all rewrite steps are semantics preserving, our approach establishes equivalence between the kernel specification and the synthesized program. This approach is implemented in the SPIRAL system, and we demonstrate it with a selection of computational kernels from the signal and image processing domain, software-defined radio, and robotic vehicle control. Our target platforms range from mobile devices, desktops, and server multicore processors to large-scale high-performance and supercomputing systems, and we demonstrate performance comparable to expertly hand-tuned code across kernels and platforms.

Linear pattern matching of compressed terms and polynomial rewriting

We consider term rewriting under sharing in the form of compression by singleton tree grammars (STG), which is more general than the term dags. Algorithms for the subtasks of rewriting are analysed: finding a redex for rewriting by locating a position for a match, performing a rewrite step by constructing the compressed result and executing a sequence of rewrite steps. The first main result is that locating a match of a linear termsin another termtcan be performed in polynomial time ifs,tare both STG-compressed. This generalizes results on matching of STG-compressed terms, matching of straight-line-program-compressed strings with character-variables, where every variable occurs at most once, and on fully compressed matching of strings. Also, for the case wheresis directed-acyclic-graph (DAG)-compressed, it is shown that submatching can be performed in polynomial time. The general case of compressed submatching can be computed in non-deterministic polynomial time, and an algorithm is described that may be exponential in the worst case, its complexity isnO(k), wherekis the number of variables with double occurrences insandnis the size of the input. The second main result is that in case there is an oracle for the redex position, a sequence ofmparallel or single-step rewriting steps under STG-compression can be performed in polynomial time. This generalizes results on DAG-compressed rewriting sequences. Combining these results implies that for an STG-compressed term rewrite system with left-linear rules,mparallel or single-step term rewrite steps can be performed in polynomial time in the input sizenandm.

Reversible computation in term rewriting

Essentially, in a reversible programming language, for each forward computation from state S to state S′, there exists a constructive method to go backwards from state S′ to state S. Besides its theoretical interest, reversible computation is a fundamental concept which is relevant in many different areas like cellular automata, bidirectional program transformation, or quantum computing, to name a few.In this work, we focus on term rewriting, a computation model that underlies most rule-based programming languages. In general, term rewriting is not reversible, even for injective functions; namely, given a rewrite step t1→t2, we do not always have a decidable method to get t1 from t2. Here, we introduce a conservative extension of term rewriting that becomes reversible. Furthermore, we also define two transformations, injectivization and inversion, to make a rewrite system reversible using standard term rewriting. We illustrate the usefulness of our transformations in the context of bidirectional program transformation.

On h-Lexicalized Restarting Automata

Following some previous studies on restarting automata, we introduce a refined model - the h-lexicalized restarting automaton (h-RLWW). We argue that this model is useful for expressing lexicalized syntax in computational linguistics. We compare the input languages, which are the languages traditionally considered in automata theory, to the so-called basic and h-proper languages, which are (implicitly) used by categorial grammars, the original tool for the description of lexicalized syntax. The basic and h-proper languages allow us to stress several nice properties of h-lexicalized restarting automata, and they are suitable for modeling the analysis by reduction and, subsequently, for the development of categories of a lexicalized syntax. Based on the fact that a two-way deterministic monotone restarting automaton can be transformed into an equivalent deterministic monotone RL-automaton in (Marcus) contextual form, we obtain a transformation from monotone RLWW-automata that recognize the class CFL of context-free languages as their input languages to deterministic monotone h-RLWW-automata that recognize CFL through their h-proper languages. Through this transformation we obtain automata with the complete correctness preserving property and an infinite hierarchy within CFL, based on the size of the read/write window. Additionally, we consider h-RLWW-automata that are allowed to perform multiple rewrite steps per cycle, and we establish another infinite hierarchy above CFL that is based on the number of rewrite steps that may be executed within a cycle. The corresponding separation results and their proofs illustrate the transparency of h-RLWW-automata that work with the (complete or cyclic) correctness preserving property

Weighted restarting automata and pushdown relations

Weighted restarting automata have been introduced to study quantitative aspects of computations of restarting automata. Here we study the special case that words over a given (output) alphabet are assigned as weights to the transitions of a restarting automaton. In this way the automaton is extended to define a mapping from the words over its input alphabet into the semiring of formal languages over a given (output) alphabet, generalizing the restarting transducers introduced by Hundeshagen (2013) [7]. We establish that the monotone restarting transducers that are allowed to use auxiliary symbols characterize the class of almost-realtime pushdown relations, and we characterize the deterministic pushdown functions by a particular type of deterministic monotone restarting transducer. Further, we show that already linearly bounded (word-)weighted monotone restarting automata that use auxiliary symbols are more expressive than the corresponding restarting transducers, both in the deterministic as well as in the nondeterministic case. Finally, we prove that for (word-)weighted monotone restarting automata with auxiliary symbols, the variant that may keep on reading after performing a rewrite step (the so-called RRWW-automaton) is strictly more expressive than the variant that must restart immediately after performing a rewrite step (the so-called RWW-automaton), which again holds in the deterministic as well as in the nondeterministic case. This is the first time that a version of the monotone RRWW-automaton is shown to differ in expressive power from the corresponding version of the monotone RWW-automaton.

Formal semantics and efficient analysis of Timed Rebeca in Real-Time Maude

The actor model is a well-established and intuitive model of distributed computation. Timed Rebeca is a timed extension of the actor-based modeling language Rebeca. Although Rebeca is supported by a rich verification toolset, Timed Rebeca has not had an executable formal semantics, and has therefore had limited support for formal analysis. In this paper, we provide a formal semantics for Timed Rebeca in Real-Time Maude. Our semantics exploits the isolation inherent in the actor model: since no actor can access the state variables of other actors, and since actors execute only one message server at a time, the effect of executing a statement is not observable by other actors. We can therefore “group together” a number of statements and execute them in one “atomic” rewrite step; this significantly improves the performance of model checking compared to standard approaches where each action is performed by a rewrite step. We have automated the translation from Timed Rebeca to Real-Time Maude, allowing Timed Rebeca models to be automatically analyzed using Real-Time Maude's reachability analysis tool and LTL and timed CTL model checkers. This enables a formal model-engineering methodology which combines the convenience of intuitive modeling in Timed Rebeca with formal verification in Real-Time Maude. We illustrate this methodology—and the performance gained by our new “partial-order-reduction-like” optimized semantics—with a number of case studies, including the IEEE 802.11 RTS/CTS collision avoidance protocol for wireless networks.

Lambda-confluence for context rewriting systems

Clearing restarting automata and limited context restarting automata are particular types of context rewriting systems. A word w is accepted by such a system if there is a sequence of rewritings that reduces the word w to the empty word λ, where each rewrite rule is extended by certain context conditions. If each rewrite step is strictly length-reducing, as for example in the case of clearing restarting automata, then the word problem for such a system can be solved nondeterministically in quadratic time. If, in addition, the contextual rewritings happen to be λ-confluent, that is, confluent on the congruence class of the empty word, then the word problem can be solved deterministically in linear time. Here we show that λ-confluence is decidable in polynomial time for limited context restarting automata of type R2, but that this property is not even recursively enumerable for clearing restarting automata. The latter follows from the fact that λ-confluence is not recursively enumerable for finite factor-erasing string-rewriting systems.

Restarting automata with auxiliary symbols restricted by lookahead size

This paper presents a study on lookahead hierarchies for restarting automata with auxiliary symbols. We show that the class of languages for deterministic monotone or monotone restarting automaton, whose restart step and rewrite step are separated, coincides with that of the same type of restarting automaton whose restart and rewrite steps are not separated, for any fixed lookahead size. For the non-monotone deterministic case, the lookahead length must be approximately doubled. We then turn our attention to restarting automata with small lookahead. For the general restarting automaton model, we show that there are just two different classes of languages recognized, through the restriction of lookahead size: those with lookahead size 1 and those with lookahead size 2. We also show that the respective (left-) monotone restarting automaton models characterize the context-free languages and that the respective right–left-monotone restarting automata characterize the linear languages both with just lookahead length 2.

Constraint-aware Schema Transformation

Data schema transformations occur in the context of software evolution, refactoring, and cross-paradigm data mappings. When constraints exist on the initial schema, these need to be transformed into constraints on the target schema. Moreover, when high-level data types are refined to lower level structures, additional target schema constraints must be introduced to balance the loss of structure and preserve semantics.We introduce an algebraic approach to schema transformation that is constraint-aware in the sense that constraints are preserved from source to target schemas and that new constraints are introduced where needed. Our approach is based on refinement theory and point-free program transformation. Data refinements are modeled as rewrite rules on types that carry point-free predicates as constraints. At each rewrite step, the predicate on the reduct is computed from the predicate on the redex. An additional rewrite system on point-free functions is used to normalize the predicates that are built up along rewrite chains.We implemented our rewrite systems in a type-safe way in the functional programming language Haskell. We demonstrate their application to constraint-aware hierarchical-relational mappings.

On stateless deterministic restarting automata

The transitions of a stateless automaton do not depend on internal states but solely on the symbols currently scanned by its head accessing the input and memory. We investigate stateless deterministic restarting automata that, after executing a rewrite step, continue to read their tape before performing a restart. Even the weakest class thus obtained contains the regular languages properly. The relations between different classes of stateless automata as well as between stateless automata and the corresponding types of automata with states are investigated, and it is shown that the language classes defined by the various types of deterministic stateless restarting automata without auxiliary symbols are anti-AFLs that are not even closed under reversal.

Transforming Outermost into Context-Sensitive Rewriting

We define two transformations from term rewriting systems (TRSs) to context-sensitive TRSs in such a way that termination of the target system implies outermost termination of the original system. In the transformation based on 'context extension', each outermost rewrite step is modeled by exactly one step in the transformed system. This transformation turns out to be complete for the class of left-linear TRSs. The second transformation is called `dynamic labeling' and results in smaller sized context-sensitive TRSs. Here each modeled step is adjoined with a small number of auxiliary steps. As a result state-of-the-art termination methods for context-sensitive rewriting become available for proving termination of outermost rewriting. Both transformations have been implemented in Jambox, making it the most successful tool in the category of outermost rewriting of the last edition of the annual termination competition.

Extending Context-Sensitivity in Term Rewriting

We propose a generalized version of context-sensitivity in term rewriting based on the notion of "forbidden patterns". The basic idea is that a rewrite step should be forbidden if the redex to be contracted has a certain shape and appears in a certain context. This shape and context is expressed through forbidden patterns. In particular we analyze the relationships among this novel approach and the commonly used notion of context-sensitivity in term rewriting, as well as the feasibility of rewriting with forbidden patterns from a computational point of view. The latter feasibility is characterized by demanding that restricting a rewrite relation yields an improved termination behaviour while still being powerful enough to compute meaningful results. Sufficient criteria for both kinds of properties in certain classes of rewrite systems with forbidden patterns are presented.

Matrix Interpretations for Proving Termination of Term Rewriting

We present a new method for automatically proving termination of term rewriting. It is based on the well-known idea of interpretation of terms where every rewrite step causes a decrease, but instead of the usual natural numbers we use vectors of natural numbers, ordered by a particular nontotal well-founded ordering. Function symbols are interpreted by linear mappings represented by matrices. This method allows us to prove termination and relative termination. A modification of the latter, in which strict steps are only allowed at the top, turns out to be helpful in combination with the dependency pair transformation. By bounding the dimension and the matrix coefficients, the search problem becomes finite. Our implementation transforms it to a Boolean satisfiability problem (SAT), to be solved by a state-of-the-art SAT solver.

Modeling Pointer Redirection as Cyclic Term-graph Rewriting

We tackle the problem of data-structure rewriting including global and local pointer redirections. Each basic rewrite step may perform three kinds of actions: (i) Local redirection, the aim of which is to redirect specific pointers determined by means of a pattern; (ii) Replacement, that may add new information to data-structures; (iii) Global redirection, which is aimed at redirecting all pointers targeting a node towards another one. We define a new framework, following the double-pushout approach, where graph rewrite rules may mix these three kinds of actions in a row. We define first the category of graphs we consider and then we define rewrite rules as pairs of graph homomorphisms of the form L←K→R. In our setting, graph K is not arbitrary, it is used to encode pointer redirection. Furthermore, pushouts do not always exist and complement pushouts, when they exist, are not unique. Despite these concerns, our definition of rewriting steps is such that a rewrite rule can always be fired, once a matching is found.

SHRINKING RESTARTING AUTOMATA

Restarting automata were introduced by Jančar et al. to model the so-called analysis by reduction. A computation of a restarting automaton consists of a sequence of cycles such that in each cycle the automaton performs exactly one rewrite step, which replaces a small part of the tape content by another, even shorter word. Here we consider a natural generalization of this model, called shrinking restarting automaton, where we only require that there exists a weight function such that each rewrite step decreases the weight of the tape content with respect to that function. While it is still unknown whether the two most general types of one-way restarting automata, the RWW-automaton and the RRWW-automaton, differ in their expressive power, we will see that the classes of languages accepted by the shrinking RWW-automaton and the shrinking RRWW-automaton coincide. As a consequence of our proof, it turns out that there exists a reduction by morphisms from the language class [Formula: see text] to the class [Formula: see text]. Further, we will see that the shrinking restarting automaton is a rather robust model of computation. Finally, we will relate shrinking RRWW-automata to finite-change automata. This will lead to some new insights into the relationships between the classes of languages characterized by (shrinking) restarting automata and some well-known time and space complexity classes.

Termination Proofs for String Rewriting Systems via Inverse Match-Bounds

Annotating a letter by a number, one can record information about its history during a rewrite derivation. In each rewrite step, numbers in the reduct are updated depending on the redex numbering. A string rewriting system is called match-bounded if there is a global upper bound to these numbers. Match-boundedness is known to be a strong sufficient criterion for both termination and preservation of regular languages. We show that the string rewriting systems whose inverse (left and right hand sides exchanged) is match-bounded, also have exceptional properties, but slightly different ones. Inverse match-bounded systems need not terminate; they effectively preserve context-free languages; their sets of normalizable strings and their sets of immortal strings are effectively regular. These languages can be used to decide the normalization, the uniform normalization, the termination and the uniform termination problem for inverse match-bounded systems. We also prove that the termination problem is decidable in linear time, and that a certain strong reachability problem is decidable, thereby solving two open problems of McNaughton’s. Like match-bounds, inverse match-bounds entail linear derivational complexity on the set of terminating strings.