Primitive sets and von Mangoldt chains: Erdős Problem #1196 and beyond

Abstract: A set of integers is primitive if no number in the set divides another. We introduce a new method for bounding Erdős sums of primitive sets, suggested from output of GPT-5.4 Pro, based on Markov chains with von Mangoldt weights. The method leads to a host of applications, yet seems to have been overlooked by the prior literature since Erdős's seminal 1935 paper. As applications, we prove two 1966 conjectures of Erdős-Sárközy-Szemerédi, on primitive sets of large numbers (#1196) and on divisibility chains (#1217). The method also provides a short proof of the Erdős Primitive Set Conjecture (#164), as well as the related claim that 2 is an ''Erdős-strong'' prime. Moreover, the method resolves a revised form of the Banks-Martin conjecture, which has long been viewed as a unifying `master theorem' for the area.

• The paper introduces a Markov chain framework leveraging von Mangoldt weights to significantly improve quantitative bounds for Erdős sums of primitive sets.
• It resolves longstanding conjectures—including the Erdős-Sárközy-Szemerédi and Erdős Primitive Set Conjectures—using rigorous probabilistic and analytic methods.
• The work bridges probabilistic approaches and combinatorial flow networks, opening pathways for extending these techniques to other poset structures and computational frameworks.

Summary Essay: Primitive Sets and von Mangoldt Chains—Resolution of Erdős Problems in Number Theory

Introduction

The paper "Primitive sets and von Mangoldt chains: Erdős Problem #1196 and beyond" (2605.00301) establishes a comprehensive, modern framework for the study of primitive sets of integers through the novel introduction of Markov chain techniques based on von Mangoldt weights. The methodology successfully addresses several longstanding conjectures in additive and multiplicative number theory, notably two prominent problems posed by Erdős, Sárközy, and Szemerédi in 1966 (#1196, #1217), the Erdős Primitive Set Conjecture (#164), and the Banks-Martin conjecture. The work systematically develops new bounds and structural results by leveraging probabilistic and analytic approaches to divisibility posets, yielding sharp quantitative improvements and establishing unexpected connections to flows, convexity, and probabilistic zeta processes.

Primitive Sets and Main Methodology

A set $A$ of positive integers is primitive if no member divides another. Classical examples include the set of primes and antichains within the divisibility poset $(\mathbb{N},\, |)$. To quantify the size of such sets, the classical Erdős sum $f(A)$ is used:

$f(A) = \sum_{a \in A} \frac{1}{a \log a}$

or more generally, through doubly harmonic weights.

The paper introduces Markov chains on the divisibility poset with transition probabilities governed by the von Mangoldt function $\Lambda(q)$, enabling random construction of divisibility chains (either upward or downward). The duality between primitive sets (antichains) and divisibility chains is exploited: any primitive set intersects a divisibility chain at most once, thus probabilistic constructions can yield sharp upper bounds on $f(A)$ for various classes of primitive sets. This methodology yields both reproofs and new proofs for several pivotal results.

Resolution of Longstanding Conjectures

Erdős-Sárközy-Szemerédi Conjecture (#1196)

The first major result, Theorem 1.1, sharpens the upper bound for Erdős sums of primitive sets within $[x, \infty)$:

$f(A) \leq 1 + O\left(\frac{1}{\log x}\right)$

for $A \subseteq [x, \infty)$. The error term is explicitly quantitative and improves the prior best bound of $f(A) \leq e^{\gamma} + o(1) \approx 1.39$.

Erdős Primitive Set Conjecture (#164)

Theorem 1.2 proves that for any primitive set $(\mathbb{N},\, |)$0,

$(\mathbb{N},\, |)$1

where $(\mathbb{N},\, |)$2 is the set of primes. This resolves Erdős’s conjecture in full generality and offers a much shorter proof via the Markov chain technique.

Banks-Martin Conjecture

Theorem 1.3 addresses the odd Banks-Martin conjecture, establishing the monotonicity of Erdős sums for primitive sets composed from odd primes:

$(\mathbb{N},\, |)$3

where $(\mathbb{N},\, |)$4 denotes a set of odd primes and $(\mathbb{N},\, |)$5 the subset with at least $(\mathbb{N},\, |)$6 prime factors from $(\mathbb{N},\, |)$7.

Erdős-strong Primes

Theorem 1.4 confirms that the prime $(\mathbb{N},\, |)$8 is "Erdős-strong," i.e., primitive sets restricted to integers with least prime factor $(\mathbb{N},\, |)$9 attain the maximal Erdős sum at the singleton set $f(A)$0.

Divisibility Chains and Density Results (#1217)

Theorem 1.6 demonstrates that any set $f(A)$1 of positive integers with positive upper doubly logarithmic density contains infinite strictly increasing divisibility chains, extending classical results of Davenport and Erdős.

Markov Chains and Flow Network Duality

The main technical innovation is the construction of downward and upward Markov chains on $f(A)$2, particularly the "von Mangoldt downward chain," where transitions are governed by the von Mangoldt function and log weights:

$f(A)$3

for $f(A)$4. The use of sub-invariant and invariant weights under these chains is central. Flow network formulations are developed as alternative but equivalent proofs, relating network cuts to antichain maximality via Stanley’s polytope and LYM-type inequalities.

A further probabilistic interpretation is given by the "zeta process," a continuous-time chain induced by exponential random variables associated with primes, yielding invariant measures and coupling divisibility zeta distributions.

Strong Numerical Results and Claims

A central numerical improvement is the bound $f(A)$5 for primitive sets in $f(A)$6, with error terms shown to be sharp up to lower order. The qualitative and quantitative improvements over prior bounds are emphasized, and several claims previously known only for odd primes are established for $f(A)$7. The monotonicity of $f(A)$8 in $f(A)$9 is proven when $f(A) = \sum_{a \in A} \frac{1}{a \log a}$0 consists of odd primes, confirming a "master theorem" for primitive set sums previously unsettled.

Additionally, the paper provides new LYM-type bounds for primitive sets, applies the methodology to Ahlswede-Khachatrian-Sárközy density results, and connects the framework to approximation problems for integer dilation.

Implications and Future Directions

The theoretical implications include the completion of a fundamental classification for primitive sets in terms of Markov chain and probabilistic flows. Practically, the methods are sufficiently general to extend to other poset structures, function fields, and permutation groups, as observed by subsequent researchers. The formalization in Lean exemplifies the integration with formal proof verification frameworks, which is increasingly relevant for rigorous analytic number theory.

Technically, the Markov chain approach allows for probabilistic sampling of divisibility structures and is poised for further extension, e.g., sharper error bounds under RH or application to "approximately primitive" sets in non-integral domains. Open questions remain regarding integer dilation approximation and generalizations to sets avoiding divisibility up to large factors, where the poset structure is more elusive.

Conclusion

This paper achieves an overview between probabilistic, analytic, and combinatorial approaches to primitive sets in number theory. The establishment of Markov chain and flow network techniques not only resolves several classic Erdős conjectures but also produces new structural and numerical bounds with strong quantitative improvements. The methodology is robust and likely to impact subsequent theoretical and computational explorations in multiplicative combinatorics, analytic number theory, and related domains.