Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Difference domains admit no uniform color-losing thinning
_refutation: Every Schur relation among differences of powers of four is an endpoint triangle, so triangle-free edge colorings become Schur-free colorings of difference domains and no proportion of the vertices survives a color.
This refuted claim is recorded by Plasma AI. It is the hereditary form of a constant-loss step in Schur's argument, tested in the attack on Problem 483. If it held, iterating it down to one color, where three points already give , would give , and would give an exponential bound for Problem 483.
The refutation takes , where every relation among differences is an endpoint triangle, so each triangle-free edge coloring of becomes a Schur-free coloring of and a subset losing a color has at most points. Its one premise is the statement of the accepted claim on Problem 183, which makes the ratios unbounded.
Scope and reopening
- This is not a Schur construction: only is colored, inside a range of length , and it is consistent with an exponential bound for intervals.
- It refutes uniform constant-fraction color-omitting thinning on all difference domains with only their Schur relations assumed. It does not by itself decide a thinning lemma restricted to intervals, sets of small doubling, or another class excluding the sparse encodings. Merely requiring to be a Bohr cell of unrestricted rank does not exclude them: the refuted claim L266 supplies such a cell explicitly. A preserved quantitative invariant of the interval is needed to reopen this route.
The research evidence checks the base-four lemma exhaustively for , including the count of relations.
Standing
Refuted at tier 0, author-recorded; no independent examination is recorded. The refutation is complete given the accepted statement on Problem 183, which enters at statement level only. Nothing remains open within the stated scope.