The $Σ$-Chain Product: A Succinct Model of Automata (De)Composition (Extended Version)
Formal Languages and Automata TheoryLogic in Computer Science
Summary
The authors study a way to build complex machines called automata by connecting smaller ones. Usually, this connection method, called the cascade product, can get very big and complicated because each part depends on all previous parts. They introduce a simpler version called the Σ-chain product, where each part only looks at the input and the part just before it, making the whole setup much smaller and easier to handle. They show that these simpler chains can represent the same kinds of machines as the full cascades and help better understand certain types of regular languages.
Authors
Roberto Borelli, Davide Bresolin, Luca Geatti, Angelo Montanari, Matteo Zavatteri
Abstract
The cascade product is a fundamental construction in automata theory, enabling hierarchical composition of automata and playing a central role in decomposition results such as the Krohn-Rhodes theorem. However, its use is limited by the exponential size required to represent cascades, which stems from the fact that each component may depend on all preceding ones, leading to exponentially large alphabets. To address this issue, we introduce the $Σ$-chain product, a restricted variant in which each component depends only on the input alphabet and the component immediately preceding it. We show that $Σ$-chains achieve linear-size representations and can be exponentially more succinct than cascades. We prove that $Σ$-chains and cascades are expressively equivalent to specific classes of automata, such as permutation-reset automata. As a consequence, we derive that a language is regular if and only if it is recognized by a $Σ$-chain of permutation-reset automata. Finally, we analyze structural properties of $Σ$-chains of reset automata, including a relation with well-known subclasses of star-free languages.