We present an algorithm that solves the 3SUM problem for n real numbers in O((n^2/log^2n) (loglog n)^{O(1)}) time, improving previous solutions by about a logarithmic factor. Our framework for shaving off two logarithmic factors can be applied to other problems, such as (median,+)-convolution/matrix multiplication and algebraic generalizations of 3SUM. We also obtain the first subquadratic results on some 3SUM-hard problems in computational geometry, for example, deciding whether (the interiors of) a constant number of simple polygons have a common intersection.

We present a collection of new results on problems related to 3SUM, including:

- The first truly subquadratic algorithm for
- computing the (min,+) convolution for monotone increasing sequences with integer values bounded by O(n),
- solving 3SUM for monotone sets in 2D with integer coordinates bounded by O(n), and
- preprocessing a binary string for histogram indexing (also called jumbled indexing).

- The first algorithm for histogram indexing for any constant alphabet size that achieves truly subquadratic preprocessing time and truly sublinear query time.
- A truly subquadratic algorithm for integer 3SUM in the case when the given set can be partitioned into n^{1-delta} clusters each covered by an interval of length n, for any constant delta > 0.
- An algorithm to preprocess any set of n integers so that subsequently 3SUM on any given subset can be solved in O(n^{13/7} polylog n) time.

- Preliminary arXiv version
- PDF talk slides
- In Proc. 47th ACM Symposium on Theory of Computing (STOC), pages 31-40, 2015

Jumbled indexing is the problem of indexing a text T for queries that ask whether there is a substring of T matching a pattern represented as a Parikh vector, i.e., the vector of frequency counts for each character. Jumbled indexing has garnered a lot of interest in the last four years. There is a naive algorithm that preprocesses all answers in O(n^2 |Sigma|) time allowing quick queries afterwards, and there is another naive algorithm that requires no preprocessing but has O(n log |Sigma|) query time. Despite a tremendous amount of effort there has been little improvement over these running times.

In this paper we provide good reason for this. We show that, under a 3SUM-hardness assumption, jumbled indexing for alphabets of size omega(1) requires Omega(n^{2-epsilon}) preprocessing time or Omega(n^{1-delta}) query time for any epsilon,delta>0. In fact, under a stronger 3SUM-hardness assumption, for any constant alphabet size r >= 3 there exist describable fixed constant epsilon_r and delta_r such that jumbled indexing requires Omega(n^{2-epsilon_r}) preprocessing time or Omega(n^{1-delta_r}) query time.

- PDF file | arXiv version
- In Proc. 41st International Colloquium on Automata, Languages, and Programming (ICALP), Lecture Notes in Computer Science, volume 8572, pages 114-125, 2014

The documents contained in this directory are included by the contributing authors as a means to ensure timely dissemination of scholarly and technical work on a non-commercial basis. Copyright and all rights therein are maintained by the authors or by other copyright holders, notwithstanding that they have offered their works here electronically. It is understood that all persons copying this information will adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.

Timothy Chan (Last updated September 2018)