Answers are grounded in this paper and cite their sources.
Kowalski and Untrau asked for a quantitative Wasserstein-metric version of the convergence, due to Kowalski and Sawin, of Kloosterman paths in C([0,1]). We give one. For the two-parameter prime ensemble, the law μ_p of the polygonal Kloosterman path and the law μ of the Kloosterman process satisfy W_1(μ_p, μ) = O(√(log log p / log p)) in the uniform metric, with explicit constants, while every path law supported on at most (p−1)² points lies at distance at least a constant times (log p)^(−1/2) from μ. The rate is therefore optimal up to the factor √(log log p). The lower bound comes from a two-sided small-ball theorem for the limit process: around every admissible half-Hölder center, including boundary coefficients and endpoints, the uniform small-ball probability has negative logarithm of exact order ε^(−2), and the upper estimate holds at every continuous center. The upper rate combines the Kloosterman sheaf, with all representation-degree dependence explicit, growing moments of dyadic Fourier blocks, and an unbiased tensor Bernstein approximation of path functionals. A direct polynomial transfer of rare events also shows that, for all sufficiently large primes (with an effective threshold), uniformly over a fixed Hölder class of centers, a proportion at least ½ exp[−c_L √(log p)] of all Kloosterman paths lies within uniform distance D_L (log p)^(−1/4) of the center.
No takes yet. Share an insight, caveat, or question.
Scott Schubr (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: