Faster Algorithms for k-Orthogonal Vectors in Low Dimension

Abstract

In the Orthogonal Vectors problem (OV), we are given two families A, B of subsets of {1,…,d}, each of size n, and the task is to decide whether there exists a pair a ∈ A and b ∈ B such that a ∩ b = ∅. Straightforward algorithms for this problem run in 𝒪(n² ⋅ d) or 𝒪(2^d ⋅ n) time, and assuming SETH, there is no 2^o(d)⋅ n^{2-ε} time algorithm that solves this problem for any constant ε > 0.
Williams (FOCS 2024) presented a 𝒪̃(1.35^d ⋅ n)-time algorithm for the problem, based on the succinct equality-rank decomposition of the disjointness matrix. In this paper, we present a combinatorial algorithm that runs in randomized time 𝒪̃(1.25^d ⋅ n). This can be improved to 𝒪(1.16^d ⋅ n) using computer-aided evaluations.
We also consider a more general k-Orthogonal Vectors problem, where given k families A_1,…,A_k of subsets of {1,…,d}, each of size n, the task is to find elements a_i ∈ A_i for every i ∈ {1,…,k} such that a₁ ∩ a₂ ∩ … ∩ a_k = ∅. We show that for every fixed k ⩾ 2, there exists ε_k > 0 such that the k-OV problem can be solved in time 𝒪(2^{(1 - ε_k)⋅d} ⋅ n). We also show that, asymptotically, this is the best we can hope for: for any ε > 0 there exists a k ⩾ 2 such that 2^{(1 - ε)⋅ d} ⋅ n^𝒪(1) time algorithm for k-Orthogonal Vectors would contradict the Set Cover Conjecture.