Experimental Evaluation of Algorithms for Computing Quasiperiods

TL;DR

This paper compares various algorithms for computing quasiperiodicity, focusing on covers and seeds, to determine their practical efficiency and suitability for large real-world datasets.

Contribution

It implements and evaluates algorithms for covers and seeds, introduces simplified versions, and assesses their performance on real-world data.

Findings

Partial covers are promising for large datasets.

Simplified algorithms improve efficiency.

Real-world data confirms effectiveness of selected algorithms.

Abstract

Quasiperiodicity is a generalization of periodicity that was introduced in the early 1990s. Since then, dozens of algorithms for computing various types of quasiperiodicity were proposed. Our work is a step towards answering the question: "Which algorithm for computing quasiperiods to choose in practice?". The central notions of quasiperiodicity are covers and seeds. We implement algorithms for computing covers and seeds in the original and in new simplified versions and compare their efficiency on various types of data. We also discuss other known types of quasiperiodicity, distinguish partial covers as currently the most promising for large real-world data, and check their effectiveness using real-world data.