Estimating the Longest Increasing Subsequence in Nearly Optimal Time

TL;DR

This paper introduces a nearly optimal sublinear time algorithm for estimating the LIS of a sequence, achieving the first sub-polynomial approximation in such time and developing novel methods like Genuine-LIS and Precision Forest.

Contribution

It presents the first sub-polynomial approximation algorithm for LIS in sublinear time and introduces innovative techniques like Genuine-LIS and Precision Forest for estimation tasks.

Findings

Achieves sub-polynomial approximation for LIS in sublinear time.

Runtime matches the lower bound of (1/) for non-trivial approximation.

Develops new methods: Genuine-LIS and Precision Forest.

Abstract

Longest Increasing Subsequence (LIS) is a fundamental statistic of a sequence, and has been studied for decades. While the LIS of a sequence of length $n$ can be computed exactly in time $O(n\log n)$, the complexity of estimating the (length of the) LIS in sublinear time, especially when LIS $\ll n$, is still open. We show that for any integer $n$ and any $\lambda = o(1)$, there exists a (randomized) non-adaptive algorithm that, given a sequence of length $n$ with LIS $\ge \lambda n$, approximates the LIS up to a factor of $1/\lambda^{o(1)}$ in $n^{o(1)} / \lambda$ time. Our algorithm improves upon prior work substantially in terms of both approximation and run-time: (i) we provide the first sub-polynomial approximation for LIS in sub-linear time; and (ii) our run-time complexity essentially matches the trivial sample complexity lower bound of $\Omega(1/\lambda)$, which is required to obtain any non-trivial approximation of the LIS. As part of our solution, we develop two novel ideas which may be of independent interest: First, we define a new Genuine-LIS problem, where each sequence element may either be genuine or corrupted. In this model, the user receives unrestricted access to actual sequence, but does not know apriori which elements are genuine. The goal is to estimate the LIS using genuine elements only, with the minimal number of "genuiness tests". The second idea, Precision Forest, enables accurate estimations for composition of general functions from "coarse" (sub-)estimates. Precision Forest essentially generalizes classical precision sampling, which works only for summations. As a central tool, the Precision Forest is initially pre-processed on a set of samples, which thereafter is repeatedly reused by multiple sub-parts of the algorithm, improving their amortized complexity.