A joint central limit theorem for inversions and two-sided descents
Let π be a uniform random permutation of n elements, let I(π) be its number of inversions and let T(π) = D(π) + D(π⁻¹) be the number of descents of π plus the number of descents of π⁻¹. We show that I and T, centred and scaled by their standard deviations, converge jointly to a pair of independent standard normal variables, with the Wasserstein distance of every unit linear combination to the standard normal law at most a constant times n^(−1/2). The same holds with T replaced by any statistic that adds a fixed function over all windows of k consecutive positions of π and of π⁻¹, provided its variance grows linearly, and the triple (I, D(π), D(π⁻¹)), standardized in the same way, converges to a standard normal vector in R³. Exactly, Cov(I, T) = (n − 1)/2. The proof replaces I by its Hoeffding projection on the point process that generates π and applies Chatterjee's normal approximation by interaction graphs, as Chatterjee and Diaconis did for T alone.
Authors
- Yitzchak Shmalo (ORCID: https://orcid.org/0000-0002-2886-3049)
Institutions
- Hebrew University of Jerusalem (IL)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-16
- DOI
- https://doi.org/10.5281/zenodo.24274300
- Primary Topic
- Probability and Statistical Research
- Type
- preprint