English

Approximability of all Boolean CSPs with linear sketches

Computational Complexity 2022-02-14 v8 Data Structures and Algorithms

Abstract

In this work we consider the approximability of Max-CSP(f)\textsf{Max-CSP}(f) in the context of sketching algorithms and completely characterize the approximability of all Boolean CSPs. Specifically, given ff, γ\gamma and β\beta we show that either (1) the (γ,β)(\gamma,\beta)-approximation version of Max-CSP(f)\textsf{Max-CSP}(f) has a linear sketching algorithm using O(logn)O(\log n) space, or (2) for every ϵ>0\epsilon > 0 the (γϵ,β+ϵ)(\gamma-\epsilon,\beta+\epsilon)-approximation version of Max-CSP(f)\textsf{Max-CSP}(f) requires Ω(n)\Omega(\sqrt{n}) space for any sketching algorithm. We also prove lower bounds against streaming algorithms for several CSPs. In particular, we recover the streaming dichotomy of [CGV20] for k=2k=2 and show streaming approximation resistance of all CSPs for which f1(1)f^{-1}(1) supports a distribution with uniform marginals. Our positive results show wider applicability of bias-based algorithms used previously by [GVV17] and [CGV20] by giving a systematic way to discover biases. Our negative results combine the Fourier analytic methods of [KKS15], which we extend to a wider class of CSPs, with a rich collection of reductions among communication complexity problems that lie at the heart of the negative results.

Keywords

Cite

@article{arxiv.2102.12351,
  title  = {Approximability of all Boolean CSPs with linear sketches},
  author = {Chi-Ning Chou and Alexander Golovnev and Madhu Sudan and Santhoshini Velusamy},
  journal= {arXiv preprint arXiv:2102.12351},
  year   = {2022}
}
R2 v1 2026-06-23T23:28:38.043Z