Solving SVP and CVP in 2^n time with Discrete Gaussian Sampling – Daniel Dadush
summary
Claims an exact CVP algorithm running in 2^n time by sampling from a shifted discrete Gaussian and combining collisions; samples require parameters bounded by the largest Gram–Schmidt length and yield enough hits to target the heaviest cosets. Uses an HKZ basis to learn last K coordinates of the closest vector, recurse on 2^K shifts to reduce dimension, and proves the sampler eventually hits a bucket with the exact closest vector. Algorithm is randomized, matches prior leading term, and leaves determinism and simplification open.
AI-written summary · about
chapters (22)
- 0:00Intro
- 0:27Lattices
- 0:44Closest Vector Problem (CVP)
- 1:02Applications of SVP & CVP
- 1:47Main Result
- 2:38Outline
- 4:13Shifted Discrete Gaussian
- 4:46Discrete Gaussian and CVP
- 6:38Approximate CVP via DGS
- 8:36Hermite-Korkine-Zolotarev Basis
- 10:10Shifted DGS Sampler
- 13:22Averaging Discrete Gaussians
- 15:06Shifted DGS Combiner
- 18:31Key Inequality
- 19:26Hope for Exact CVP
- 20:08Clustering approx. closest vectors
- 21:54How many closest vectors?
- 22:54Dimension Reduction via Clustering
- 28:56High Level Algorithm
- 30:07Key Challenges
- 32:22Conclusions
- 33:49Open Problems
related talks
19mProvable Sieving Algorithms for SVP and CVP in the l_p norm
01h 13mAlgorithms for Lattice Problems
01h 06mHardness of the Shortest Vector Problem: A Simplified Proof and a Survey
59mSome Recent Results Showing that Lattice Problems Are (sort of) Equivalent in All Norms
01h 03mComplexity of Lattice Problems
01h 08mThe Mathematics of Lattices I
01h 00mA Reverse Minkowski Theorem
01h 36mAlgorithms for Algebraic Lattices: Classical and Quantum
