Abstract:Consider the following gap cycle counting problem in the streaming model: The edges of a 2-regular $n$-vertex graph $G$ are arriving one-by-one in a stream and we are promised that $G$ is a disjoint union of either $k$-cycles or $2k$-cycles for some small $k$; the goal is to distinguish between these two cases using a limited memory. Verbin and Yu [SODA 2011] introduced this problem and showed that any single-pass streaming algorithm solving it requires $n^{1-\Omega(1/k)}$ space. This result and the proof technique behind it—the Boolean Hidden Hypermatching communication problem—has since been used extensively for proving streaming lower bounds for various problems, including approximating MAX-CUT, matching size, property testing, matrix rank and Schatten norms, streaming unique games and CSPs, and many others. Despite its significance and broad range of applications, the lower bound technique of Verbin and Yu comes with a key weakness that is also inherited by all subsequent results: the Boolean Hidden Hypermatching problem is hard only if there is exactly one round of communication and, in fact, can be solved with logarithmic communication in two rounds. Therefore, all streaming lower bounds derived from this problem only hold for single-pass algorithms. Our goal in this paper is to remedy this state-of-affairs. We prove the first multi-pass lower bound for the gap cycle counting problem: Any $p$-pass streaming algorithm that can distinguish between disjoint union of $k$-cycles vs $2k$-cycles—or even $k$-cycles vs one Hamiltonian cycle-requires $n^{1-1/k^{\Omega(1/p)}}$ space. This makes progress on multiple open questions in this line of research dating back to the work of Verbin and Yu. As a corollary of this result and by simple (or even no) modification of prior reductions, we can extend many of previous lower bounds to multi-pass algorithms. For instance, we can now prove that any streaming algorithm that ($1+\varepsilon$) -approximates the value of MAX-CUT, maximum matching size, or rank of an $n$-by-$n$ matrix, requires either $n^{\Omega(1)}$ space or $\Omega(\log({}^{1}\!/\!_{\varepsilon}))$ passes. For all these problems, prior work left open the possibility of even an $O(\log n)$ space algorithm in only two passes.

A Cut-Matching Game for Constant-Hop Expanders

Online matching games in bipartite expanders and applications

New Structures and Algorithms for Length-Constrained Expander Decompositions

Maximum Length-Constrained Flows and Disjoint Paths: Distributed, Deterministic and Fast

A Faster Combinatorial Algorithm for Maximum Bipartite Matching

Expander Decomposition in Dynamic Streams

Regularized Box-Simplex Games and Dynamic Decremental Bipartite Matching

The Cutting Plane Method is Polynomial for Perfect Matchings

Partition Expanders

Maximum Bipartite Matching in $n^{2+o(1)}$ Time via a Combinatorial Algorithm

New Explicit Constant-Degree Lossless Expanders

Almost Ramanujan Expanders from Arbitrary Expanders via Operator Amplification

Towards a Unified Theory of Sparsification for Matching Problems

Multi-Pass Graph Streaming Lower Bounds for Cycle Counting, MAX-CUT, Matching Size, and Other Problems

Max-cut and Extendability of Matchings in Distance-Regular Graphs

Chernoff Bounds and Reverse Hypercontractivity on HDX

Deterministic Expander Routing: Faster and More Versatile

Cut-Matching Games on Directed Graphs

A Tight Max-Flow Min-Cut Duality Theorem for Non-Linear Multicommodity Flows

Expander Hierarchies for Normalized Cuts on Graphs

Distributed Algorithms for Maintaining Dynamic Expander Graphs