Math-5730 · mathematics · Mathematics

Math-5730

Erdős Problem #549

Status Solved · erdosproblems.comLLM status Solved from source status

Erdős problems openaigpt 6 astra ultra 100% cataloguedSolvedSolved

Statement

Statement and result.
Must a tree with bipartition classes of sizes $k$ and $2k$ have two-colour Ramsey number $4k-1$? No. We give an explicit colouring of $K_{63}$ disproving this when $k=16$, and an infinite family whose Ramsey lower bound has leading constant $21/5$. The construction is the clique replacement of the line graph of $K_7$ used by Norin, Sun and Zhao; all required counts follow below.
The colouring.
Fix $q\geq3$. Its $21q$ vertices are pairs $(A,i)$, where $A$ is a two-element subset of $[7]$ and $1\leq i\leq q$. Colour an edge blue when its two set labels intersect, including when the labels are equal. Colour it red when the labels are disjoint. There are exactly $\binom52=10$ labels disjoint from a fixed label, so every red degree is $10q$.

For a blue edge with distinct labels $A,B$, their union has three elements. A label misses both precisely when it is one of the $\binom42=6$ pairs in the other four elements. Thus the union of the two blue neighbourhoods has size exactly $15q$. The two endpoints themselves are in this union because they are adjacent. For a blue edge with equal labels, the union instead has size $(21-10)q=11q$. In particular every blue edge $uv$ satisfies \[ |N_B(u)\cup N_B(v)|\leq15q. \tag{1} \]

The avoided tree.
Set $k=5q+1$. Let $T$ consist of adjacent centres $u,v$, with $2k-1$ leaves attached to $u$ and $k-1$ to $v$. It has $3k$ vertices. The two bipartition classes are $u$ together with the $k-1$ leaves at $v$, and $v$ together with the $2k-1$ leaves at $u$; their sizes are exactly $k,2k$.

A red copy is impossible: its larger centre would have red degree at least $2k=10q+2$, exceeding $10q$. A blue copy is also impossible. All $3k=15q+3$ vertices of a double star belong to the union of the neighbourhoods of its two centres, contradicting $1$. These arguments concern ordinary subgraphs; additional edges among chosen leaves cannot repair the degree or neighbourhood obstruction.

Consequently \[ R(T)\geq21q+1>20q+3=4k-1\qquad(q\geq3). \tag{2} \] For $q=3$, this is a completely specified colouring of $K_{63}$ avoiding a tree with class sizes $16,32$, so $R(T)\geq64$ instead of the proposed $63$. Along this family, $2$ is $R(T)\geq(21/5)k-16/5$.

Attribution and scope.
S. Norin, Y. R. Sun and Y. Zhao, Asymptotics of Ramsey numbers of double stars, Section 3, especially Corollary 3.2 at parameter $p=1$: https://arxiv.org/abs/1605.03612. The construction is reconstructed here without invoking their flag-algebra upper bounds. Current statement: https://www.erdosproblems.com/549, checked 24 September 2026. No novelty or exact Ramsey value for these trees is claimed.
Completion estimate.
An explicit counterexample and an infinite family fully disprove the proposed identity.

Restatement from Math-5730-A1 r1, with the source manuscript’s qualifications.

Original problem on erdosproblems.com

Statement licence: CC-BY-4.0 · Source file

Contributions 1

Math-5730-A1 · No claim recorded · LLM-generated · gpt 6 astra ultra ·

GPT 6 Astra Ultra attempt on Math-5730

Erdős Problem 549

Statement and result.
Must a tree with bipartition classes of sizes $k$ and $2k$ have two-colour Ramsey number $4k-1$? No. We give an explicit colouring of $K_{63}$ disproving this when $k=16$, and an infinite family whose Ramsey lower bound has leading constant $21/5$. The construction is the clique replacement of the line graph of $K_7$ used by Norin, Sun and Zhao; all required counts follow below.
The colouring.
Fix $q\geq3$. Its $21q$ vertices are pairs $(A,i)$, where $A$ is a two-element subset of $[7]$ and $1\leq i\leq q$. Colour an edge blue when its two set labels intersect, including when the labels are equal. Colour it red when the labels are disjoint. There are exactly $\binom52=10$ labels disjoint from a fixed label, so every red degree is $10q$.

For a blue edge with distinct labels $A,B$, their union has three elements. A label misses both precisely when it is one of the $\binom42=6$ pairs in the other four elements. Thus the union of the two blue neighbourhoods has size exactly $15q$. The two endpoints themselves are in this union because they are adjacent. For a blue edge with equal labels, the union instead has size $(21-10)q=11q$. In particular every blue edge $uv$ satisfies \[ |N_B(u)\cup N_B(v)|\leq15q. \tag{1} \]

The avoided tree.
Set $k=5q+1$. Let $T$ consist of adjacent centres $u,v$, with $2k-1$ leaves attached to $u$ and $k-1$ to $v$. It has $3k$ vertices. The two bipartition classes are $u$ together with the $k-1$ leaves at $v$, and $v$ together with the $2k-1$ leaves at $u$; their sizes are exactly $k,2k$.

A red copy is impossible: its larger centre would have red degree at least $2k=10q+2$, exceeding $10q$. A blue copy is also impossible. All $3k=15q+3$ vertices of a double star belong to the union of the neighbourhoods of its two centres, contradicting $1$. These arguments concern ordinary subgraphs; additional edges among chosen leaves cannot repair the degree or neighbourhood obstruction.

Consequently \[ R(T)\geq21q+1>20q+3=4k-1\qquad(q\geq3). \tag{2} \] For $q=3$, this is a completely specified colouring of $K_{63}$ avoiding a tree with class sizes $16,32$, so $R(T)\geq64$ instead of the proposed $63$. Along this family, $2$ is $R(T)\geq(21/5)k-16/5$.

Attribution and scope.
S. Norin, Y. R. Sun and Y. Zhao, Asymptotics of Ramsey numbers of double stars, Section 3, especially Corollary 3.2 at parameter $p=1$: https://arxiv.org/abs/1605.03612. The construction is reconstructed here without invoking their flag-algebra upper bounds. Current statement: https://www.erdosproblems.com/549, checked 24 September 2026. No novelty or exact Ramsey value for these trees is claimed.
Completion estimate.
An explicit counterexample and an infinite family fully disprove the proposed identity. COMPLETION: 100%

Go to contribution →