On the Complexity of Coordinated Table Selective Substitution Systems

TL;DR

This paper analyzes the computational complexity of languages generated by coordinated table selective substitution systems, showing their acceptance by Turing machines within logarithmic space and time bounds, and relating them to Petri nets and one-counter languages.

Contribution

It establishes complexity bounds for languages generated by specific substitution systems and connects these classes to well-known computational complexity classes.

Findings

RL; 0S)-systems are accepted by nondeterministic TMs in O(log n) space and O(nlog n) time.

(RL; RB)-systems are accepted by ATMs in O(log n) time and space.

These classes are included in NSPACE(log n) and SPACE(log n), respectively.

Abstract

We investigate computational resources used by Turing machines (TMs) and alternating Turing machines (ATMs) to accept languages generated by coordinated table selective substitution systems with two components. We prove that the class of languages generated by real-time (RL; 0S)-systems, an alternative device to generate lambda-free labeled marked Petri nets languages, can be accepted by nondeterministic TMs in O(log n) space and O(nlog n) time. Consequently, this proper sub-class of Petri nets languages (known also as L-languages) is included in NSPACE(log n). The class of languages generated by (RL; RB)-systems for which the nonterminal alphabet of the RL-grammar is composed of only one symbol and the nonterminal alphabet of the RB-grammar is composed of two symbols, can be accepted by ATMs in O(log n) time and space. Consequently, this proper subclass of one-counter languages generated by one-counter machines with only one control state is included in U_E*-uniform NC1, hence in SPACE(log n).