Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated

Sums of Proper Divisors with Missing Digits

../

theorem_1_8: Gives a quantitative density-zero bound for inputs whose sum of proper divisors uses only digits from a fixed proper subset in base g at least three.


Kübra Benli, Giulia Cesana, Cécile Dartyge, Charlotte Dombrowsky, and Lola Thompson, Sums of Proper Divisors with Missing Digits, arXiv:2307.12859v1 (24 July 2023). The work was subsequently published in Research Directions in Number Theory, Association for Women in Mathematics Series 32 (Springer, 2024), 93--110; those publication details do not identify the bytes selected here as the published edition.

Local artifact. The selected 14-page arXiv v1 PDF was supplied by the reviewer; the exact historical acquisition time of these bytes was not recorded and remains unknown. The arXiv record (https://arxiv.org/abs/2307.12859, read 2026-10-02) names the Creative Commons Attribution 4.0 license.

The paper studies the preimage under the sum-of-proper-divisors function s(n)=σ(n)−ns(n)=\sigma(n)-n of sets whose members use only a prescribed proper subset of the base-gg digits. Theorem 1.8 fixes g≥3g\geq3, γ∈(0,1)\gamma\in(0,1), and a nonempty proper digit set D⊊{0,1,…,g−1}D\subsetneq\{0,1,\ldots,g-1\}, and proves

#{n≤x: every base-g digit of s(n) lies in D}=O ⁣(xexp⁡(−(log⁡log⁡x)γ)).\#\{n\leq x:\text{ every base-}g\text{ digit of }s(n)\text{ lies in }D\} =O\!\left(x\exp\bigl(-(\log\log x)^\gamma\bigr)\right).

Because the target digit set has asymptotic density zero, this verifies the EGPS preimage conjecture for this structured class. It does not settle the conjecture for an arbitrary density-zero target set. The authors note that when 1∈D1\in D, prime inputs give a lower bound π(x)\pi(x) because s(p)=1s(p)=1. Thus digit sets containing 11 show that the logarithmic exponent cannot be uniformly improved over the full class of fixed gg and DD; this is not a matching lower bound for every fixed digit set DD.

Section 4 proves Theorem 1.8. The main split is according to whether a chosen power gkg^k divides σ(n)\sigma(n): Lemma 1.9 controls the nondivisible case, and in the divisible case the digit restriction confines nn to at most ∣D∣k|D|^k residue classes modulo gkg^k. This is a proof pointer, not a complete proof.

Bears on. #955.

Results to transcribe.

  • Theorem 1.8: the quantitative missing-digit preimage bound for g≥3g\geq3.

Living verification. Needs review. The identity, selected version, Theorem 1.8 statement, special-case transfer, and proof route were checked against the selected arXiv v1 PDF. No complete proof is supplied, reconstructed, or independently certified here.