Skip to content

Releases: ritabe-dev/ErdosProblem160-OneThirdUpper

E160 one-third upper bound - unreviewed candidate v0.2.1

Choose a tag to compare

This prerelease hardens the review artifact for the candidate upper-bound improvement for Erdős Problem 160. No Lean theorem, assumption, constant, or proof was changed.

Result

Let H(N) be the least number of colours needed so that every nontrivial four-term arithmetic progression in {1, ..., N} uses at least three colours. The manuscript gives the candidate bound

H(N) <= N^(1/3 + o(1)),

and the Lean development checks the stated theorem chain, including the eventual form H(N) <= N^(1/3 + epsilon) for every real epsilon > 0. This would improve the exponent log(3)/log(22) currently recorded on the maintained Erdős Problems page. It is an upper-bound partial result and does not solve Problem 160.

Changes since v0.2.0

  • clarifies that the total Lean definition has the unused endpoint value siteH 0 = 0, while all stated bounds concern positive N;
  • pins Tectonic 0.16.9, bundle v33, and the paper build timestamp, and checks a byte-identical PDF rebuild on Ubuntu 24.04;
  • checks the exact names and count of the nine audited theorem endpoints;
  • scans all first-party Lean sources for placeholders and conventional standalone primitive declarations;
  • adds a one-file release manifest and verifies the remote tag, tagged PDF, and GitHub Release asset together.

Verification

The manuscript has not been peer reviewed. No claim of priority or independently established novelty is made.

E160 one-third upper bound - unreviewed candidate v0.2.0

Choose a tag to compare

This prerelease contains a proof manuscript and Lean 4 formalization of the
stated theorem chain for a candidate upper-bound improvement for Erdős Problem
160.

Result

Let H(N) be the least number of colours needed so that every nontrivial
four-term arithmetic progression in {1, ..., N} uses at least three colours.
The manuscript proves

H(N) <= N^(1/3 + o(1)).

More precisely, for every real epsilon > 0, it proves the eventual bound

H(N) <= N^(1/3 + epsilon).

This would improve the exponent log(3)/log(22) currently recorded on the
maintained Erdős Problems page. It is an upper-bound partial result and does
not solve Problem 160. The manuscript has not been peer reviewed, and no claim
of priority is made.

Contents

  • the proof manuscript, TeX source, and compiled PDF;
  • a Lean 4 formalization of the stated theorem chain and source-convention bridge;
  • pinned Lean and mathlib versions with a one-command verification check;
  • a theorem map, source-statement audit, and literature search.

Reproduce the formal artifact with:

bash scripts/check_release.sh

Tag v0.2.0-review-candidate points to root commit
42604f944f7367e13e9ae43072bded87930864df. Both the main-branch and tagged
GitHub Actions verification runs passed for this tree.

PDF SHA-256:

703081d94eebaceda64f4e90cc79aa0007994b874e44f66c46231c36984ab74d