Several types of systems of parallel communicating restarting automata are introduced and studied. The main result shows that, for all types of restarting automata X, centralized systems of restarting automata of type X have the same computational power as non-centralized systems of restarting automata of the same type and the same number of components. This result is proved by a direct simulation. In addition, it is shown that for one-way restarting automata without auxiliary symbols, systems of degree at least two are more powerful than the component automata themselves. Finally, a lower and an upper bound are given for the expressive power of systems of parallel communicating restarting automata.

Based on the formal framework of reaction systems by Ehrenfeucht and Rozenberg [Fund. Inform. 75 (2007) 263–280], reaction automata (RAs) have been introduced by Okubo et al. [Theoret. Comput. Sci. 429 (2012) 247–257], as language acceptors with multiset rewriting mechanism. In this paper, we continue the investigation of RAs with a focus on the two manners of rule application: maximally parallel and sequential. Considering restrictions on the workspace and the λ-input mode, we introduce the corresponding variants of RAs and investigate their computation powers. In order to explore Turing machines (TMs) that correspond to RAs, we also introduce a new variant of TMs with restricted workspace, called s(n)-restricted TMs. The main results include the following: (i) for a language L and a function s(n), L is accepted by an s(n)-bounded RA with λ-input mode in sequential manner if and only if L is accepted by a log s(n)-bounded one-way TM; (ii) if a language L is accepted by a linear-bounded RA in sequential manner, then L is also accepted by a P automaton [Csuhaj−Varju and Vaszil, vol. 2597 of Lect. Notes Comput. Sci. Springer (2003) 219–233.] in sequential manner; (iii) the class of languages accepted by linear-bounded RAs in maximally parallel manner is incomparable to the class of languages accepted by RAs in sequential manner.

We introduce and investigate several variants of a bidirectional string assembling system, which is a computational model that generates strings from copies of assembly units. The underlying mechanism is based on two-sided piecewise assembly of a double-stranded sequence of symbols, where the upper and lower strand have to match. The generative capacities and the relative power of the variants are our main interest. In particular, we prove that bidirectional string assembling system generate languages not represented as any finite concatenation of one-sided string assembling systems. The latter build an infinite, strict and tight concatenation hierarchy. Moreover, it is shown that even the strongest system in question can only generate NL languages, while there are unary regular languages that cannot be derived. Furthermore, a finite strict hierarchy with respect to the different variants considered is shown and closure properties of the languages generated are presented.

In the literature various types of restarting automata have been studied that are based on contextual rewriting. A word w is accepted by such an automaton if, starting from the initial configuration that corresponds to input w, the word w is reduced to the empty word by a finite number of applications of these contextual rewritings. This approach is reminiscent of the notion of McNaughton families of languages. Here we put the aforementioned types of restarting automata into the context of McNaughton families of languages, relating the classes of languages accepted by these automata in particular to the class GCSL of growing context-sensitive languages and to the class CRL of Church–Rosser languages.

The recently introduced model of transducing by observing is compared with traditional models for computing transductions on the one hand and the recently introduced restarting transducers on the other hand. Most noteworthy, transducing observer systems with length-reducing rules are almost equivalent to RRWW-transducers. With painter rules we obtain a larger class of relations that additionally includes nearly all rational relations.

Linear finite transducers underlie a series of schemes for Public Key Cryptography (PKC) proposed in the 90s of the last century. The uninspiring and arid language then used, condemned these works to oblivion. Although some of these schemes were afterwards shown to be insecure, the promise of a new system of PKC relying on different complexity assumptions is still quite exciting. The algorithms there used depend heavily on the results of invertibility of linear transducers. In this paper we introduce the notion of post-initial linear transducer, which is an extension of the notion of linear finite transducer with memory, and for which the previous fundamental results on invertibility still hold. This extension enabled us to give a new method to obtain a left inverse of any invertible linear finite transducer with memory. It also plays an essencial role in the necessary and sufficient condition that we give for left invertibility of linear finite transducers.

We consider conditional tabled Lindenmayer sytems without interaction, where each table is associated with a regular set and a table can only be applied to a sentential form which is contained in its associated regular set. We study the effect to the generative power, if we use instead of arbitrary regular languages only finite, nilpotent, monoidal, combinational, definite, ordered, union-free, star-free, strictly locally testable, commutative regular, circular regular, and suffix-closed regular languages. Essentially, we prove that the hierarchy of language families obtained from conditional Lindenmayer systems with subregular conditions is almost identical to the hierarchy of families of subregular languages.