Sensitivity of string compressors and repetitiveness measures
Abstract
The sensitivity of a string compression algorithm asks how much the output size for an input string can increase when a single character edit operation is performed on . This notion enables one to measure the robustness of compression algorithms in terms of errors and/or dynamic changes occurring in the input string. In this paper, we analyze the worst-case multiplicative sensitivity of string compression algorithms, which is defined by , where denotes the edit distance between and . For the most common versions of the Lempel-Ziv 77 compressors, we prove that the worst-case multiplicative sensitivity is upper bounded by a small constant, and give matching lower bounds. We generalize these results to the smallest bidirectional scheme . In addition, we show that the sensitivity of a grammar-based compressor called GCIS is also a small constant. Further, we extend the notion of the worst-case sensitivity to string repetitiveness measures such as the smallest string attractor size and the substring complexity , and show that the worst-case sensitivity of is also a small constant. These results contrast with the previously known related results such that the size of the Lempel-Ziv 78 factorization can increase by a factor of [Lagarde and Perifel, 2018], and the number of runs in the Burrows-Wheeler transform can increase by a factor of [Giuliani et al., 2021] when a character is prepended to an input string of length . By applying our sensitivity bounds of or the smallest grammar to known results (c.f. [Navarro, 2021]), some non-trivial upper bounds for the sensitivities of important string compressors and repetitiveness measures including , , LZ-End, RePair, LongestMatch, and AVL-grammar are derived.
Keywords
Cite
@article{arxiv.2107.08615,
title = {Sensitivity of string compressors and repetitiveness measures},
author = {Tooru Akagi and Mitsuru Funakoshi and Shunsuke Inenaga},
journal= {arXiv preprint arXiv:2107.08615},
year = {2023}
}
Comments
The journal version is superceded by this version (we added almost tight bounds for Bisection with insertions and deletions)