LIPIcs.ICALP.2021.119.pdf
- Filesize: 0.76 MB
- 17 pages
In this work, we revisit the problem of testing membership in regular languages, first studied by Alon et al. [Alon et al., 2001]. We develop a one-sided error property tester for regular languages under weighted edit distance that makes 𝒪(ε^{-1} log(1/ε)) non-adaptive queries, assuming that the language is described by an automaton of constant size. Moreover, we show a matching lower bound, essentially closing the problem for the edit distance. As an application, we improve the space bound of the current best streaming property testing algorithm for visibly pushdown languages from 𝒪(ε^{-4} log⁶ n) to 𝒪(ε^{-3} log⁵ n log log n), where n is the size of the input. Finally, we provide a Ω(max(ε^{-1}, log n)) lower bound on the memory necessary to test visibly pushdown languages in the streaming model, significantly narrowing the gap between the known bounds.
Feedback for Dagstuhl Publishing