Faster Recovery of Approximate Periods over Edit Distance

TL;DR

This paper presents a significantly faster algorithm for approximate period recovery in words, reducing the complexity from O(n^{4/3}) to O(n log n) by leveraging a new approach to approximate pattern matching in periodic texts.

Contribution

It introduces an improved algorithm for approximate period recovery, enhancing efficiency from previous methods by applying a novel technique for pattern matching in periodic texts.

Findings

Reduced time complexity from O(n^{4/3}) to O(n log n)

Efficient verification for candidate periods given as cyclic rotations

Applicable to approximate pattern matching in periodic texts

Abstract

The approximate period recovery problem asks to compute all $\textit{approximate word-periods}$ of a given word $S$ of length $n$: all primitive words $P$ ($|P|=p$) which have a periodic extension at edit distance smaller than $\tau_p$ from $S$, where $\tau_p = \lfloor \frac{n}{(3.75+\epsilon)\cdot p} \rfloor$ for some $\epsilon>0$. Here, the set of periodic extensions of $P$ consists of all finite prefixes of $P^\infty$. We improve the time complexity of the fastest known algorithm for this problem of Amir et al. [Theor. Comput. Sci., 2018] from $O(n^{4/3})$ to $O(n \log n)$. Our tool is a fast algorithm for Approximate Pattern Matching in Periodic Text. We consider only verification for the period recovery problem when the candidate approximate word-period $P$ is explicitly given up to cyclic rotation; the algorithm of Amir et al. reduces the general problem in $O(n)$ time to a logarithmic number of such more specific instances.