Low moments of sums of extended Rademacher random multiplicative functions
Christopher Atherfold, Maxim Gerspach, Jad Hamdan
math.NT
Nov 3, 2026 · v1
math.PR
TL;DR
All results were formalised in Lean 4, with Claude generating the proofs, using Mathlib and the PNT+ project; axiom checks confirm nothing is left unproved.
Abstract
Let $X^e \colon \mathbb{N} \to \{\pm 1\}$ be the random, completely multiplicative function whose values at the primes $(X^e(p))_{p \text{ prime}}$ are independent and uniformly distributed on $\{ \pm 1 \}$. Let $X=X^eμ^2$. We determine the order of magnitude of \[ \mathbb{E} \Bigg[ \bigg| \frac{1}{\sqrt{x}}\sum_{n \le x} X^e(n) \bigg|^{2 q} \Bigg]\quad \text{ and }\quad \mathbb{E} \Bigg[ \bigg|\sum_{n \le x} \frac{X(n)}{\sqrt{n}} \bigg|^{2 q} \Bigg] \] for $q\in [0,1]$, showing that a transition from better-than-square-root cancellation to growth like a power of $\log x$ occurs when $q$ crosses $1/2$. In particular, the first absolute moments are of order $(\log\log x)^{3/4}$. The proof proceeds by comparing these sums to integrals of random Euler products, which are then studied by analogy with Gaussian multiplicative chaos integrals. The main difficulty lies in proving the lower bound at $q=1/2$, for which we also adapt arguments from the study of random walks.
Problem
The paper determines the order of magnitude of low moments (2q-th moments, q in [0,1]) of partial sums of extended Rademacher random multiplicative functions, and of weighted sums of Rademacher random multiplicative functions. These sums model random quadratic character sums.
Approach
The sums are compared with integrals of random Euler products. Those integrals are analysed by analogy with Gaussian multiplicative chaos. Random walk arguments are adapted to prove the lower bound at q=1/2. All results were formalised in Lean 4 with Claude, building on Mathlib and the PrimeNumberTheoremAnd (PNT+) project. A build-time check confirms that each result depends only on Lean's standard axioms.
Results
The moments are of order (log log x)^{-q/2} for q<1/2, (log log x)^{3/4} at q=1/2, and (log x)^{q(2q-1)} for q>1/2. This shows a transition from better-than-square-root cancellation to power-of-log growth. The full formalisation contains no unproved steps.
| Range of q | Moment order |
|---|
| 0 ≤ q < 1/2 | (log log x)^{-q/2} |
| q = 1/2 | (log log x)^{3/4} |
| 1/2 < q ≤ 1 | (log x)^{q(2q-1)} |
Order of magnitude of the 2q-th moments (both sums)