Abstract. A language L over a finite alphabet Σ is called growth-sensitive if forbidding any non-empty set F of subwords yields a sub-language L F whose exponential growth rate is smaller than that of L. Say that a context-free grammar (and associated language) is ergodic if its dependency di-graph is strongly connected. It is known that regular and unambiguous non-linear context-free languages which are ergodic are growth-sensitive. In this note it is shown that ergodic unambiguous linear languags are growth-sensitive, closing the gap that remained open.
IntroductionLet L be a language over the alphabet Σ, that is, a subset of the free monoid Σ * of all finite words over Σ. We write ε for the empty word and Σ + = Σ * \ {ε}. For a word w ∈ Σ * , its length (number of letters) is denoted by |w|. The growth rate of L ⊆ Σ * is the numberAlso recall that L has exponential growth if γ(L) > 1 and it has sub-exponential growth otherwise, namely, if γ(L) = 1. A language L over Σ is growth-sensitive iffor any non-empty F ⊂ Σ * consisting of subwords of elements of L, where L F = {w ∈ L : no v ∈ F is a subword of w}.A context-free grammar is a quadruple C = (V, Σ, P, S), where V is a finite set of variables, disjoint from the finite alphabet Σ, the variable S is the start symbol, and P ⊂ V × (V ∪ Σ) * is a finite set of production rules. We write T u or (T u) ∈ P if (T, u) ∈ P. For v, w ∈ (V ∪ Σ) * , we write v =⇒ w if v = v 1 T v 2 and w = v 1 uv 2 , whereReceived by the editors November 4, 2004.2000 Mathematics Subject Classification. Primary 68Q45; Secondary 05A16, 05C20, 68Q42. Key words and phrases. Context-free grammar, linear language, dependency di-graph, ergodicity, ambiguity, growth, higher block languages, automaton, bilateral automaton.
605License or copyright restrictions may apply to redistribution; see https://www.ams.org/journal-terms-of-use
TULLIO CECCHERINI-SILBERSTEINA context-free language is a language generated by a context-free grammar. If in addition the production rules T u are such that u contains at most one variable in V, then the grammar and the corresponding language are termed linear. In particular, if u ∈ Σ * ∪ Σ * V (resp. Σ * ∪ VΣ * ), then the language is left (resp. right) linear. In both cases, language and grammar are also called regular.A general context-free grammar C is called unambiguous if for any w ∈ L(C) there is a unique rightmost derivation S * =⇒ w. A context-free language is unambiguous if it is generated by some unambiguous grammar. Note that there are context-free languages that cannot be generated by unambiguous grammars: these are called inherently ambiguous languages. In our setting, the languageis linear and inherently ambiguous (using Ogden's iteration lemma [17] (see also Chapter 6 in [9]) it can be deduced that one always has two different derivations for the words of the form a n b n c n ). Thus there exist inherently ambiguous linear languages.We shall always assume to have a reduced grammar C, that is, each variable is used in some rightmost derivation of a wor...