3-Uniform Generalised Hedgehogs in Ramsey Theory

• 3-uniform generalised hedgehogs are hypergraphs defined by partitioning vertices into a body and spikes, where each spike attaches uniquely to a body-pair.
• They are central in hypergraph Ramsey theory, with techniques like peeling and degeneracy orders used to derive two-colour Ramsey bounds.
• Their Ramsey numbers exhibit non-linear behavior, with bounds of O(n^(3/2)) and Ω(n^(3/2)/log n), challenging classical Burr–Erdős conjectures.

Searching arXiv for papers on 3-uniform generalised hedgehogs and related Ramsey results. arXiv search query: "3-uniform generalised hedgehogs Ramsey numbers hedgehogs" A 3-uniform generalised hedgehog is a 3-graph $H_n=(V,E)$ equipped with a distinguished partition $V=B\sqcup S$, where $B$ is the body, $S$ is the set of spikes, and there is a surjection $f:S\to \binom{B}{2}$ such that

$E=\bigl\{\{s\}\cup f(s): s\in S\bigr\}.$

Thus each spike belongs to exactly one 3-edge, formed with a body-pair, and there are no other edges. This family extends the standard 3-uniform hedgehog by allowing multiple spikes to attach to the same pair of body vertices. The class has become a central test case in hypergraph Ramsey theory because it simultaneously exhibits sparse local structure and unexpectedly large two-colour Ramsey numbers, while its standard subfamily already displays strong dependence on the number of colours (Allen et al., 31 Jul 2025, Conlon et al., 2015).

1. Definition and formal structure

In the formulation studied by Allen–Boyadzhiyska–Pavez-Signes, a 3-uniform generalised hedgehog with body size $|B|=n$ consists of a body $B$, a spike set $S$, and a surjection

$f:S\longrightarrow \binom{B}{2},$

with edge set

$V=B\sqcup S$0

Equivalently, every spike $V=B\sqcup S$1 forms exactly one triple $V=B\sqcup S$2 with a pair $V=B\sqcup S$3, and there are no other edges. The standard hedgehog is the special case in which $V=B\sqcup S$4 and $V=B\sqcup S$5 is a bijection; the generalised version requires only $V=B\sqcup S$6 and permits several spikes to map to the same body-pair (Allen et al., 31 Jul 2025).

This definition isolates a highly structured form of sparsity. All edges contain exactly one spike and two body vertices, so the entire hypergraph is determined by the attachment pattern of spikes to body-pairs. A plausible implication is that the Ramsey problem is governed less by global density than by the combinatorics of how frequently a given body-pair must be completed to a monochromatic triple.

2. The standard hedgehog as a special case

The classical 3-uniform hedgehog $V=B\sqcup S$7 is defined on

$V=B\sqcup S$8

with edge set

$V=B\sqcup S$9

Its body is $B$0, and each unordered pair of body vertices is extended by exactly one spine vertex. The total number of vertices is

$B$1

and the total number of edges is

$B$2

Every induced subhypergraph has a vertex of degree at most $B$3, so the standard hedgehog is 1-degenerate (Conlon et al., 2015).

The paper of Conlon, Fox, and Rödl uses the notation $B$4 for the same standard object, with

$B$5

It also records a $B$6-uniform analogue $B$7, defined on $B$8 vertices by requiring each $B$9-subset of the first $S$0 vertices to extend in exactly one way to a $S$1-edge (Fox et al., 2019).

The distinction between standard and generalised hedgehogs is decisive. In the standard case, each body-pair is represented exactly once; in the generalised case, some body-pairs may carry many spikes, while others need only be represented to satisfy surjectivity. That relaxation is precisely what underlies the larger two-colour Ramsey numbers discovered later.

3. Two-colour Ramsey numbers

For the standard hedgehog, the first general upper bound established that

$S$2

Since the total number of vertices of $S$3 is $S$4, this is linear in the body size and polynomial in the number of vertices (Conlon et al., 2015).

Conlon, Fox, and Rödl asked whether the two-colour Ramsey number of the hedgehog is nearly linear in the total number of vertices. They answered this affirmatively by proving that if $S$5 and

$S$6

then every red/blue colouring of $S$7 contains a monochromatic $S$8. In particular,

$S$9

Because $f:S\to \binom{B}{2}$0, this is nearly linear in the number of vertices (Fox et al., 2019).

The behaviour changes for 3-uniform generalised hedgehogs. Allen–Boyadzhiyska–Pavez-Signes prove that there is an absolute constant $f:S\to \binom{B}{2}$1 such that for any two generalised hedgehogs $f:S\to \binom{B}{2}$2,

$f:S\to \binom{B}{2}$3

and in the proof sketch obtain the explicit form

$f:S\to \binom{B}{2}$4

They also show that this order is essentially best possible by constructing a generalised hedgehog with

$f:S\to \binom{B}{2}$5

so the standard hedgehog’s near-linear two-colour behaviour does not extend to the full generalised class (Allen et al., 31 Jul 2025).

4. The refined proof for the standard hedgehog

The $f:S\to \binom{B}{2}$6 bound for the standard hedgehog is obtained by combining a greedy “peeling” procedure with a terminal dichotomy. One begins with a working vertex set $f:S\to \binom{B}{2}$7, initially all vertices, and for each $f:S\to \binom{B}{2}$8 and integer $f:S\to \binom{B}{2}$9 defines

$E=\bigl\{\{s\}\cup f(s): s\in S\bigr\}.$0

This set may be viewed as the blue-heavy neighbours of $E=\bigl\{\{s\}\cup f(s): s\in S\bigr\}.$1. In Stage $E=\bigl\{\{s\}\cup f(s): s\in S\bigr\}.$2, for $E=\bigl\{\{s\}\cup f(s): s\in S\bigr\}.$3 from $E=\bigl\{\{s\}\cup f(s): s\in S\bigr\}.$4 up to $E=\bigl\{\{s\}\cup f(s): s\in S\bigr\}.$5, if some $E=\bigl\{\{s\}\cup f(s): s\in S\bigr\}.$6 satisfies $E=\bigl\{\{s\}\cup f(s): s\in S\bigr\}.$7, one peels $E=\bigl\{\{s\}\cup f(s): s\in S\bigr\}.$8 into a growing blue body set $E=\bigl\{\{s\}\cup f(s): s\in S\bigr\}.$9. Fractional penalties $|B|=n$0 are imposed on would-be spines so that the cumulative deletion cost remains controlled (Fox et al., 2019).

Two outcomes are possible. If peeling fails to reach $|B|=n$1, then it gets stuck only after removing at most

$|B|=n$2

vertices, and the remaining colouring is balanced in the sense that

$|B|=n$3

for all $|B|=n$4 and $|B|=n$5. A separate probabilistic lemma shows that any such balanced colouring on $|B|=n$6 vertices already contains a monochromatic hedgehog. If peeling succeeds and builds a body $|B|=n$7, then a Hall-matching argument assigns to each body-pair $|B|=n$8 a distinct spine vertex $|B|=n$9 such that $B$0 is blue. The key condition is

$B$1

together with a degree-splitting lemma ensuring that each final body-pair lies in at least $B$2 monochromatic triples of the chosen colour (Fox et al., 2019).

This argument is specific to the standard hedgehog’s rigid one-spine-per-pair geometry. It shows that the standard object is far more Ramsey-friendly than generalised variants with highly nonuniform multiplicities.

5. Upper-bound architecture for generalised hedgehogs

The $B$3 upper bound for 3-uniform generalised hedgehogs proceeds through an auxiliary 2-graph $B$4 built from a red/blue colouring of the complete 3-graph $B$5. A pair $B$6 is coloured red in $B$7 if it lies in fewer than $B$8 red triples of $B$9, and blue in $S$0 if it lies in fewer than $S$1 blue triples. A lemma extracted from Conlon–Fox–Rödl states that every vertex $S$2 can be marked either red or blue so that $S$3 has at most $S$4 neighbours in one of the two colours in $S$5; if $S$6 had more than $S$7 red-neighbours and more than $S$8 blue-neighbours, counting triples through $S$9 would give a contradiction (Allen et al., 31 Jul 2025).

Consequently, at least half of the vertices are sparse in one colour; by symmetry there is a red set $f:S\longrightarrow \binom{B}{2},$0 of size at least $f:S\longrightarrow \binom{B}{2},$1. To embed a red copy of a target generalised hedgehog $f:S\longrightarrow \binom{B}{2},$2, one forms a graph $f:S\longrightarrow \binom{B}{2},$3 on the body $f:S\longrightarrow \binom{B}{2},$4, where the edges are exactly the body-pairs that occur in some spike-edge of $f:S\longrightarrow \binom{B}{2},$5. In the proof sketch,

$f:S\longrightarrow \binom{B}{2},$6

so the degeneracy $f:S\longrightarrow \binom{B}{2},$7 of $f:S\longrightarrow \binom{B}{2},$8 satisfies

$f:S\longrightarrow \binom{B}{2},$9

The body vertices are then placed in a $V=B\sqcup S$00-degeneracy order. At each step, already embedded neighbours in $V=B\sqcup S$01 rule out only their red-neighbours in $V=B\sqcup S$02, and each such set has size at most $V=B\sqcup S$03, so a fresh image remains available in $V=B\sqcup S$04. After the body is embedded, each body-pair lies in at least $V=B\sqcup S$05 red triples of $V=B\sqcup S$06, and the spikes can be attached greedily using fresh vertices (Allen et al., 31 Jul 2025).

This proof replaces the standard hedgehog’s peeling-and-matching strategy with a degeneracy-guided body embedding. The change reflects a structural shift: once several spikes may attach to the same body-pair, the central issue is no longer a perfect matching between body-pairs and distinct spine candidates, but rather controlling the dependency graph among the body-pairs that actually matter.

6. Lower bounds, colour sensitivity, and the Burr–Erdős context

The lower bound for generalised hedgehogs is obtained from a random graph gadget. One first constructs a graph $V=B\sqcup S$07 satisfying three simultaneous properties: every vertex has degree at most $V=B\sqcup S$08, $V=B\sqcup S$09 contains no $V=B\sqcup S$10, and $V=B\sqcup S$11 has no independent set of size $V=B\sqcup S$12. One then colours each triple $V=B\sqcup S$13 red if at least one of $V=B\sqcup S$14 is an edge of $V=B\sqcup S$15, and blue otherwise. The target hypergraph $V=B\sqcup S$16 is chosen so that its body has size $V=B\sqcup S$17, each body-pair has exactly one spike attached, and among the first $V=B\sqcup S$18 body vertices each of the $V=B\sqcup S$19 pairs receives $V=B\sqcup S$20 extra spikes, after which the vertex set is filled up arbitrarily to total size $V=B\sqcup S$21. A blue copy would yield an independent set of size $V=B\sqcup S$22, while a red copy would force a $V=B\sqcup S$23, because the heavy pairs among the first $V=B\sqcup S$24 body vertices have more than $V=B\sqcup S$25 red triples each. This yields the lower bound

$V=B\sqcup S$26

showing that the $V=B\sqcup S$27 upper bound is near-sharp (Allen et al., 31 Jul 2025).

The earlier literature had already established that hedgehogs are highly sensitive to the number of colours. For the standard 3-uniform hedgehog,

$V=B\sqcup S$28

so the four-colour Ramsey number grows exponentially. Conlon, Fox, and Rödl further showed that for $V=B\sqcup S$29 colours the Ramsey number is exponential in $V=B\sqcup S$30, while for $V=B\sqcup S$31 it is superpolynomial, with lower bound

$V=B\sqcup S$32

Thus the two-colour case is the only one exhibiting nearly linear behaviour for the standard hedgehog (Conlon et al., 2015, Fox et al., 2019).

These results place 3-uniform generalised hedgehogs at the center of the hypergraph Burr–Erdős program. Allen–Boyadzhiyska–Pavez-Signes show that for $V=B\sqcup S$33 there exists a 1-degenerate 3-graph $V=B\sqcup S$34 with

$V=B\sqcup S$35

so the hypergraph Burr–Erdős conjecture already fails in the two-colour, uniformity-3 setting. Combined with the $V=B\sqcup S$36 upper bound, this completely settles the two-colour, 3-uniform case: every 1-degenerate 3-graph has Ramsey number between $V=B\sqcup S$37 and $V=B\sqcup S$38, and no linear bound is possible (Allen et al., 31 Jul 2025).