CryptoRecords.info
SVP / Classical · worst case

One Discrete Gaussian Sample in 2^{n/2+o(n)} Time

Jiseung Kim

arXiv 2608.03220

PreprintSource checked
Time bound
2c₀ n + o(n), c₀ ≈ 0.7315
Space bound
2n/2 + o(n)
Source / version date
4 Aug 2026
Comparison variant
Classical · worst case
Cost scale
2c n + o(n)
Norm
Euclidean ℓ₂
Success guarantee
At least 2/3
Computation model
Randomized Turing machine; polynomial input-length factors suppressed
reference
Theorem 6.6; §6 constants

Assumptions and source notes

The exact-SVP consequence is separate from the paper’s one-sample DGS theorem. Theorem 6.6 gives success probability at least 2/3, time coefficient c₀ = 1/2 + β_Lev²/(4e ln 2) ≈ 0.7315 and space coefficient 1/2 on every execution. This was not an overall time record: Hhan’s 0.6039 result had appeared the preceding day. Source checked; proof not independently audited.

Source location

Abstract; §6 constant definitions, printed p. 17; Theorem 6.6, printed p. 23
Version: arXiv:2608.03220v1, 4 August 2026; original PDF checked

Selected earlier references

Selected sources cited by this paper.

  1. Solving the Shortest Vector Problem in 2^n Time via Discrete Gaussian Sampling

    Divesh Aggarwal · Daniel Dadush · Oded Regev · Noah Stephens-Davidowitz · 2015

    Uses the older above-smoothing batch sampler on superlattices to obtain one arbitrary-width centered sample. Abstract; §1.2; Theorem 2.6

  2. Solving the Shortest Vector Problem in 2^(0.7314n+o(n)) Time via Discrete Gaussian Sampling on Superlattices

    Yiming Gao · Yansong Feng · Honggang Hu · 2026

    Independent superlattice-based exact-SVP result; compares 0.7314 with this paper’s 0.7315. §1, Concurrent work; bibliography [GFH26]

  3. Solving the Shortest Vector Problem in 2^(0.6039n) Time via Mid-point Hessian

    Minki Hhan · 2026

    Earlier Hessian approach already gives a smaller exact-SVP time exponent. §1, Concurrent work; bibliography [Hha26]

  4. Solving the Shortest Lattice Vector Problem in Time 2^(2.465n)

    Xavier Pujol · Damien Stehlé · 2009

    Uses the earlier lattice-point counting bound in the Gaussian-mass analysis. §6, Lemma 6.1 and bibliography [PS09]

Primary source ↗

@misc{kim-2026-one-dgs,
  author = {Jiseung Kim},
  title = {{One Discrete Gaussian Sample in 2^{n/2+o(n)} Time}},
  year = {2026},
  url = {https://arxiv.org/abs/2608.03220},
  cr_date = {2026-08-04},
  cr_status = {preprint},
  cr_author_ids = {jiseung-kim},
  archiveprefix = {arXiv},
  eprint = {2608.03220}
}