English

Sparse Convolution for Approximate Sparse Instance

Data Structures and Algorithms 2023-06-06 v1

Abstract

Computing the convolution ABA \star B of two vectors of dimension nn is one of the most important computational primitives in many fields. For the non-negative convolution scenario, the classical solution is to leverage the Fast Fourier Transform whose time complexity is O(nlogn)O(n \log n). However, the vectors AA and BB could be very sparse and we can exploit such property to accelerate the computation to obtain the result. In this paper, we show that when ABc1=k\|A \star B\|_{\geq c_1} = k and ABc2=nk\|A \star B\|_{\leq c_2} = n-k holds, we can approximately recover the all index in suppc1(AB)\mathrm{supp}_{\geq c_1}(A \star B) with point-wise error of o(1)o(1) in O(klog(n)log(k)log(k/δ))O(k \log (n) \log(k)\log(k/\delta)) time. We further show that we can iteratively correct the error and recover all index in suppc1(AB)\mathrm{supp}_{\geq c_1}(A \star B) correctly in O(klog(n)log2(k)(log(1/δ)+loglog(k)))O(k \log(n) \log^2(k) (\log(1/\delta) + \log\log(k))) time.

Cite

@article{arxiv.2306.02381,
  title  = {Sparse Convolution for Approximate Sparse Instance},
  author = {Xiaoxiao Li and Zhao Song and Guangyi Zhang},
  journal= {arXiv preprint arXiv:2306.02381},
  year   = {2023}
}