Cover
Every unguarded vertex touches a guard. This is ordinary domination.
Solved theorem · connected induced-P₅-free graphs · α ≥ 3
How many guards does a network need if every guard must be able to answer an attack without leaving the rest exposed? A published upper bound was 3/2 times a natural graph parameter. The sharp multiplier is smaller: 4/3.
γs is the fewest secure guards. α is the largest group of mutually nonadjacent vertices. The statement assumes α ≥ 3.
First, the game
Every unguarded vertex touches a guard. This is ordinary domination.
When an unguarded vertex is attacked, one adjacent guard swaps into it.
After that swap, every vertex must still be covered. The smallest team that always can do this is γs.
What does P₅-free mean? The network contains no five-vertex route whose only links are the four consecutive ones. That missing pattern forces surprisingly strong global structure.
Interactive proof map
Why 4/3 is exact
An exhaustive verifier checks all 220 triples and finds no secure one; 435 four-sets are secure.
Explore the 12-vertex graph ↗For α ≥ 3, α + 1 ≤ 4α/3. So every graph stays at or below the witness ratio.
Read the formal assembly ↗How the proof was found
The 12-vertex witness disproved γs ≤ α, but exposed the sharp lower ratio 4/3.
Cut vertices, dominating pairs, and dominating three-vertex paths were proved to satisfy γs ≤ α + 1.
Triangles were closed first. Larger minimal dominating cliques split into private regions and multi-hub neighbours.
A 12-vertex SAT witness killed the claim that every dominating edge extends directly to a secure triple. The proof target was changed instead of forced.
A private-region edge forces one hub plus its two endpoints to dominate the graph. That drops the final case back into the solved path branch.
Evidence boundary
The theorem is an all-orders mathematical argument. Three independent proof reviews reconstructed its crucial cross-edge case. Finite computation is supporting evidence, not the logical foundation.
Solved—and not claimed beyond scope