A Complete Taxonomy of Restarting Automata without Auxiliary Symbols*

A complete taxonomy is presented for restarting automata without auxiliary symbols. In this taxonomy, the language classes that are accepted by deterministic and nondeterministic, monotone, weakly monotone, and non-monotone, shrinking and length-reducing restarting automata are compared to each other with respect to inclusion. As it turns out, the 45 types of restarting automata considered yield 29 different classes of languages. By presenting a collection of rather simple example languages, it is shown that, for any two of these language classes ℒ1 and ℒ2, the class ℒ1 is a subclass of ℒ2 if and only if ℒ1 is defined by a type of restarting automaton that is a restriction of a type of restarting automaton that defines the class ℒ2.

Automata with cyclic move operations for picture languages

Here, we study the cyclic extensions of Sgraffito automata and of deterministic two-dimensional two-way ordered restarting automata for picture languages. Such a cyclically extended automaton can move in a single step from the last column (or row) of a picture to the first column (or row). For Sgraffito automata, we show that this cyclic extension does not increase the expressive power of the model, while for deterministic two-dimensional two-way restarting automata, the expressive power is strictly increased by allowing cyclic moves. In fact, for the latter automata, we take the number of allowed cyclic moves in any column or row as a parameter, and we show that already with a single cyclic move per column (or row) the deterministic two-dimensional extended two-way restarting automaton can be simulated. On the other hand, we show that two cyclic moves per column or row already give the same expressive power as any finite number of cyclic moves.

ON THE DESCRIPTIONAL COMPLEXITY OF THE WINDOW SIZE FOR DELETING RESTARTING AUTOMATA

The restarting automaton was inspired by the technique of ‘analysis by reduction’ from linguistics. A restarting automaton processes a given input word through a sequence of cycles. In each cycle the current word on the tape is scanned from left to right and a single local simplification (a rewrite) is executed. One of the essential parameters of a restarting automaton is the size of its read/write window. Here we study the impact of the window size on the descriptional complexity of several types of deterministic and nondeterministic restarting automata. For all k ≥ 4, we show that the savings in the economy of descriptions of restarting automata that can only delete symbols but not rewrite them (that is, the so-called R- and RR-automata) cannot be bounded by any recursive function, when changing from window size k to window size k + 1. This holds for deterministic as well as for nondeterministic automata, and for k ≥ 5, it even holds for the stateless variants of these automata. However, the trade-off between window sizes two and one is recursive for deterministic devices. In addition, a polynomial upper bound is given for the trade-off between RRWW-automata with window sizes k + 1 and k for all k ≥ 2.

COOPERATING DISTRIBUTED SYSTEMS OF RESTARTING AUTOMATA

Here we introduce cooperating distributed systems of restarting automata and establish that in mode = 1 they correspond to the non-forgetting restarting automaton.

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.