Alphabeta Math
TheoremStatement: Literature-sourcedProof: AI-generatedverified 2026-09-24 (gpt-6-sol)
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

Deleting a leaf and a co-leaf preserves virality of a finite forbidden family

Statement

Let F be a finite family of finite graphs. Let H1∈F have a leaf v, and let H2∈F have a co-leaf w. Write H1′=H1−v and H2′=H2−w. If

F1={H1′}∪(F∖{H1}),F2={H2′}∪(F∖{H2})

are both viral, then F is viral.

Facts & Assumptions

Given: The family, chosen graphs and vertices, and viral modified families in the statement.

[L1]

A co-leaf of H2 is a leaf of H2‾ (Co-leaves of a finite graph). Complementation preserves labelled induced embedding counts: ind⁡H2(G)=ind⁡H2‾(G‾), and the same holds after deleting w.

[L2]

For each fixed graph H and 0<η<1/2, Nikiforov's theorem gives δ>0 such that fewer than (δ∣G∣)∣H∣ induced H-embeddings force an η-restricted set of size at least δ∣G∣ (Nikiforov: for every H and every ϵ∈(0,12) there is δ>0 such that every graph G with ind⁡H(G)<(δ∣V(G)∣)∣V(H)∣ has an ϵ-restricted vertex set of size at least δ∣V(G)∣).

[L3]

The leaf-extension blockade lemma supplies its three quantitative outcomes for ordinary graphs with a leaf (A sparse host has many leaf extensions, few smaller copies, or a long sparse blockade). Applying it in G‾ to the leaf of H2‾ gives the corresponding dense-blockade alternative in G.

Proof

technique · iterative sparsification and a divisive blockade
1.1

If ∣H1∣≤2 or ∣H2∣≤2, the chosen graph has two vertices: a leaf or co-leaf cannot occur on fewer. Every graph on two vertices has the Erdős-Hajnal property by the at-most-three-vertex theorem, so its singleton family is viral by the single-graph equivalence. The same viral exponent works for F: the few-copy premise for F includes the premise for that singleton. Hence assume both orders are at least three. Put h=max⁡{∣H1∣,∣H2∣,4},c=4−h.

givenL1
1.2

Apply [L2] to H1 with η=c2, obtaining δ>0; shrink δ to at most 1 if necessary. Choose a common viral exponent d≥4 for F1,F2 large enough that cd≤δ. Define a=2d2h,b=a+6d+1. For any 0<x<c, if every member of F has fewer than (xb∣G∣)∣H∣ copies in G, then [L2] supplies a c2-restricted set of size at least δ∣G∣≥cd∣G∣, because xb<cb≤cd≤δ.

L2givenalgebra
1.3

We establish a one-step assertion. Let 0<x≤y≤c, and let Q be a y2-restricted graph on q vertices. Then at least one of these happens:

  • some H∈F has more than (xb−4dq)∣H∣ copies in Q;
  • for i=1 or 2, some S⊆V(Q) has ∣S∣≥y2q and ind⁡Hi′(Q[S])<(y2d2∣S∣)∣Hi′∣;
  • Q has an x-sparse or (1−x)-dense blockade of length at least y−1 and width at least ya+1q.

If Q has maximum degree at most y2q, apply [L3] to H1 in Q. Its first outcome is stronger than the first here because 2a+2∣H1∣≤(b−4d)∣H1∣; its second is stronger than the second here because a−2>2d2(∣H1∣−1) and y<1; and its third is the sparse blockade above. If Q‾ has maximum degree at most y2q, apply [L3] to H2‾ in Q‾ and translate the counts and blockade back using [L1]. Since Q is restricted, one of these cases applies. [L1, L3, step 1.2, algebra]

2.1

We prove that (b,c) witnesses divisiveness. Fix 0<x<c and a nonempty graph G on n≥x−b/2 vertices satisfying the defining few-copy bounds for F. Suppose it has no blockade required by divisiveness. Let m≥2 be the least integer with cdm−1≤x. We construct nested sets V(G)=S0⊇S1⊇⋯⊇Sm such that, for 1≤i≤m, |S_i|\ge c^{3d^i}|S_{i-1}|, \quad G[S_i]\text{ is }c^{2d^{i-1}}\text{-restricted}.\tag{1} For i=1, step 1.2 gives a c2-restricted set of size at least cdn≥c3dn.

step 1.2L2
3.1

Suppose (1) is constructed through Si with 1≤i<m, and put y=cdi−1. Minimality of m gives x<cdm−2≤y; also y≤c. Summing the geometric exponents in (1), with d≥4, gives |S_i|\ge c^{3(d+\cdots+d^i)}n\ge c^{4d^i}n =y^{4d}n\ge x^{4d}n.\tag{2} Consequently every H∈F has fewer than (xb−4d∣Si∣)∣H∣ copies in G[Si], because it has fewer than (xbn)∣H∣ copies in G. Apply step 1.3 to Q=G[Si]. Its first outcome is impossible. Its blockade outcome would have length at least y−1∈[2,x−1] and width at least ya+1∣Si∣≥ya+4d+1n≥ybn because b=a+6d+1; it would be the forbidden divisive blockade. Thus its second outcome supplies S⊆Si of size ∣S∣≥y2∣Si∣ with few Hj′ copies for some j∈{1,2}.

step 1.3step 2.1algebra
4.1

Since (2) gives ∣S∣≥y4d+2n≥y5dn, we have y2d2∣S∣≥y2d2+5dn≥xbn. The final inequality uses x≤y<1 and b=a+6d+1≥2d2+5d. Thus every unchanged member of Fj has fewer than (y2d2∣S∣)∣H∣ copies; the replaced member Hj′ has the same strict bound by step 3.1. Apply virality of Fj with parameter y2d to obtain Si+1⊆S that is y2d=c2di-restricted and has ∣Si+1∣≥y2d2∣S∣≥y2d2+2∣Si∣≥c3di+1∣Si∣. The last inequality follows from 2d2+2≤3d2. This completes the induction.

step 1.2step 3.1algebra
5.1

At the final stage, cdm−1≤x, so G[Sm] is x2-restricted. The geometric sum in (1), together with the minimality inequality x<cdm−2, yields |S_m|\ge c^{4d^m}n\ge x^{4d^2}n\ge x^{-1},\tag{3} where the last step uses n≥x−b/2 and b≥8d2+2. Put k=⌈x−1/2⌉ and r=⌊2x∣Sm∣⌋. As x<c≤1/16, one has 2≤k≤x−1, kr≤4x1/2∣Sm∣≤∣Sm∣, and r≥x∣Sm∣≥1. Choose k disjoint r-element subsets of Sm. If G[Sm] is sparse, they form an x-sparse blockade: each later vertex has at most x2∣Sm∣≤xr neighbors in an earlier block. If G‾[Sm] is sparse, they form the dense alternative. Finally r≥x∣Sm∣≥x4d2+1n≥n/kb, since k≥x−1/2 and b≥8d2+2. This contradicts step 2.1. Hence F is divisive, and [L4] makes it viral.

step 2.1step 4.1L4algebra∎

Source notes

The source's ordered Theorem 6.1 has one leaf and one leaf in the complement; its overbars matter. The proof above applies the two leaf counting lemmas directly to ordinary induced embeddings. Their counting and blockade arguments do not require order, while complementing the second graph and host handles the dense case.

Depends on

Used by

Dependency tree · two levels

37 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.

Sources