Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Statement
The first proof defines numbers by the recurrence
with the initial values , , (2), and states: "We obtain e. g. easily
The function of the introduction is (4). For the paper gives a graph-theoretic formulation and "a very simple proof" (p. 466): Theorem. "In an arbitrary graph let the maximum number of independent points be ; if the number of points is then there exists in our graph a complete graph of order ." (p. 466, footnote marks omitted; the proof's base case counts an edge as a complete graph of order 1.)
As a statement about Ramsey numbers, (3) with the Theorem gives , that is and . The paper's is defined by the recurrence, not as the least Ramsey number, so (3) is an identity for that function and an upper bound for the Ramsey number. The paper proves no lower bound for any Ramsey number.
Source. P. Erdős and G. Szekeres, A combinatorial problem in geometry, Compositio Mathematica 2 (1935), 463--470; (1)--(4) and the Theorem on printed p. 466 (PDF p. 4 of the scan), read on the page image; the proof of the Theorem is on p. 467, with the counting step (5) .
Read depth. Claims checked: (1), (2), (3), (4) and the Theorem were read clause by clause on the page image. The inductive proof of (1) (pp. 464--466) and of the Theorem (p. 467) were not checked.
Proof pointer
The recurrence (1) comes from the induction on pp. 464--466 over the two-class colorings of -element subsets; (3) is the solution of (1), (2) for (Pascal's rule). The graph Theorem is stated on p. 466 and proved on p. 467 by induction on : one of the points of a maximum independent set is joined to at least others, among which the induction hypothesis gives a complete graph of order once (5); with that point it forms one of order .
Dependencies
None outside the paper.
Bears on
- Problem 1029: the classical upper bound on the diagonal Ramsey number; the site's commentary attributes both and the upper bound to this paper, but the lower bound is Erdős's 1947 probabilistic bound, which the paper does not contain.
- Problem 986: the upper bound for fixed that the problem's lower bound matches up to a polylogarithmic factor.
- Problem 77: the source of the upper end in , since ; the first exponential improvement below came in 2023.