LPF Computation Revisited

摘要

We present efficient algorithms for storing past segments of a text. They are computed using two previously computed read-only arrays (SUF and LCP) composing the Suffix Array of the text. They compute the maximal length of the previous factor (subword) occurring at each position of the text in a table called LPF. This notion is central both in many conservative text compression techniques and in the most efficient algorithms for detecting motifs and repetitions occurring in a text.rnThe main results are: a linear-time algorithm that computes explicitly the permutation that transforms the LCP table into the LPF table; a time-space optimal computation of the LPF table; and an O(n log n) strong in-place computation of the LPF table.