29
10
v1v2 (latest)

Improved Hardness of BDD and SVP Under Gap-(S)ETH

Abstract

We show improved fine-grained hardness of two key lattice problems in the p\ell_p norm: Bounded Distance Decoding to within an α\alpha factor of the minimum distance (BDDp,α\mathrm{BDD}_{p, \alpha}) and the (decisional) γ\gamma-approximate Shortest Vector Problem (SVPp,γ\mathrm{SVP}_{p,\gamma}), assuming variants of the Gap (Strong) Exponential Time Hypothesis (Gap-(S)ETH). Specifically, we show: 1. For all p[1,)p \in [1, \infty), there is no 2o(n)2^{o(n)}-time algorithm for BDDp,α\mathrm{BDD}_{p, \alpha} for any constant α>αkn\alpha > \alpha_\mathsf{kn}, where αkn=2ckn<0.98491\alpha_\mathsf{kn} = 2^{-c_\mathsf{kn}} < 0.98491 and cknc_\mathsf{kn} is the 2\ell_2 kissing-number constant, unless non-uniform Gap-ETH is false. 2. For all p[1,)p \in [1, \infty), there is no 2o(n)2^{o(n)}-time algorithm for BDDp,α\mathrm{BDD}_{p, \alpha} for any constant α>αp\alpha > \alpha^\ddagger_p, where αp\alpha^\ddagger_p is explicit and satisfies αp=1\alpha^\ddagger_p = 1 for 1p21 \leq p \leq 2, αp<1\alpha^\ddagger_p < 1 for all p>2p > 2, and αp1/2\alpha^\ddagger_p \to 1/2 as pp \to \infty, unless randomized Gap-ETH is false. 3. For all p[1,)2Zp \in [1, \infty) \setminus 2 \mathbb{Z} and all C>1C > 1, there is no 2n/C2^{n/C}-time algorithm for BDDp,α\mathrm{BDD}_{p, \alpha} for any constant α>αp,C\alpha > \alpha^\dagger_{p, C}, where αp,C\alpha^\dagger_{p, C} is explicit and satisfies αp,C1\alpha^\dagger_{p, C} \to 1 as CC \to \infty for any fixed p[1,)p \in [1, \infty), unless non-uniform Gap-SETH is false. 4. For all p>p02.1397p > p_0 \approx 2.1397, p2Zp \notin 2\mathbb{Z}, and all C>CpC > C_p, there is no 2n/C2^{n/C}-time algorithm for SVPp,γ\mathrm{SVP}_{p, \gamma} for some constant γ>1\gamma > 1, where Cp>1C_p > 1 is explicit and satisfies Cp1C_p \to 1 as pp \to \infty, unless randomized Gap-SETH is false.

View on arXiv
Comments on this paper