slug: fuentes-2025-mmi-forbidden-subgraph bibkey: fuentes2025mmigraphstates doi: 10.48550/arXiv.2511.19585 url: https://arxiv.org/abs/2511.19585v1 triage: theorem motivation_gids:
- D5/S3/Quantum/Entanglement/GraphStateMonogamyForbiddenSubgraph.result
The forbidden-subgraph conjecture for MMI in graph states holds
Problem
Fuentes, Keeler, Munizzi and Pollack (arXiv:2511.19585v1) conjecture:
Conjecture 1 (Forbidden-Subgraph). Any graph state that violates MMI has a graph representation that is -equivalent to a graph containing a subgraph. is with respect to some partition.
Issue #10534 fixes the readings from the paper: the entropy of a set of
qubits A is the rank over Z_2 of the adjacency block with rows in A and
columns outside A; an MMI violation is a triple of pairwise disjoint sets
I, J, K with S_IJ + S_IK + S_JK < S_I + S_J + S_K + S_IJK; LC
equivalence is a finite sequence of local complementations; K_4 is the
four-vertex star K_{1,3}, read as an induced subgraph, which is the stronger
reading; and a generalized star with centre C and parts V^1, …, V^k has
nonempty, pairwise disjoint parts that with C partition the vertices, no
edge between different parts, and an edge from every part to C.
Motivation
The paper verifies the conjecture exhaustively up to eight qubits and leaves
larger graph states open.
D5/S3/Quantum/Entanglement/GraphStateMonogamyForbiddenSubgraph.result
proves it for every number of qubits: every MMI-violating graph state reaches,
by local complementations, a graph with an induced four-star, which is a
generalized star for the partition that puts the three leaves in their own
parts and every other vertex in the centre.
Gap
Issue #10534 preregisters the proof route and the literature check. arXiv
lists only version 1; the three citing papers reported by Semantic Scholar and
INSPIRE (arXiv:2608.17020, arXiv:2604.14271, arXiv:2510.15067) cite the paper
only in reference lists, and the authors’ 2026 papers do not address the
conjecture. Kwon and Oum (Eur. J. Combin. 41 (2014) 100–127, arXiv:1306.3066,
Theorem 3.1(2)) prove that every connected graph with enough vertices has a
K_n vertex-minor, with an unspecified bound; K_4 and the four-vertex star
are locally equivalent. The explicit bound of seven vertices and the list of
the connected graphs without a four-star vertex-minor used here are
not-found-in-searched-scope.
Route
- Local complementation at vertices of a vertex set commutes with restriction to that set, so an induced four-star, or a relabelling of a fixed graph, reached on part of the vertices lifts to the whole graph.
- A connected graph has an ordering of its vertices in which every vertex
after the first is adjacent to an earlier one. The graph on the first
m + 1vertices is the graph on the firstmwith one new vertex adjacent to a nonempty setTof them, and local complementation at earlier vertices keepsTnonempty. For the representativesK_1,K_2, the pathP_3, the pathP_4, the 5-cycleC_5and the triangular prism, and every nonemptyT, a listed sequence of at most three local complementations carries the extension to a graph with an induced four-star or to a relabelling of the next representative; every extension of the prism reaches an induced four-star. These 120 certificates are checked by evaluation. Hence a connected graph without a four-star vertex-minor has at most six vertices and is locally equivalent to a relabelled representative. - Local complementation at
vchanges the adjacency block ofAby the row operation1 + u e_vᵀ(u_v = 0) whenv ∈ A, an involution overZ_2, and by the corresponding column operation otherwise, so every entropy is unchanged; relabelling permutes rows and columns. - When a vertex set has no edge to its complement, the adjacency block of
every
Ais block-diagonal after reordering, so the entropies add over the set and its complement. - The representatives satisfy MMI: for
K_1,K_2,P_3,C_5and the prism the entropy ofAismin(|A|, n − |A|), by a checked parity criterion that makes the rows ofAindependent, and MMI then reduces to arithmetic; forP_4a violation needs four nonempty parts, hence single vertices, where the three pairs have entropies at least2, 2, 1. - If a graph state reached no induced four-star, every connected component would be locally equivalent to a relabelled representative (steps 1–2), so by steps 3–5 and induction on the number of vertices it would satisfy MMI.
Falsifier
The proof would fail if a certificate did not check, if local complementation changed some entropy, or if a representative violated MMI.
Evidence
Exhaustive search with nauty geng over the connected graphs on 4 to 8
vertices: the local-complementation orbits without an induced four-star are
the orbits of P_4 (n = 4), C_5 (n = 5) and the triangular prism
(n = 6), and there are none for n = 7 (853 graphs) and n = 8 (11117
graphs). No graph in these orbits violates MMI. The 5-cycle and the prism
give the absolutely maximally entangled states on five and six qubits.
The canonical source is
D5/S3/Quantum/Entanglement/GraphStateMonogamyForbiddenSubgraph.lean.
Its public declarations are lc, lcSeq, IsClaw, entropy,
ViolatesMMI, IsGeneralizedStar, claim, and result. The frozen module
state has statement identity
sha256:81981077a3cd56f484109eb63008d9318f6a36c1c70e0df494f9f09515aef166.
The result declaration has statement identity
sha256:5685717741e6283d7382bd2db3f27e7d30d268b94ec809b0114eb4a1b0bdc566.
The Freeze event is
sha256:327c0570404b9ac857e7ddd982af73a180197bbbc816d647781967cc4de43657.
It has no project-level frozen prerequisite. The proof uses only the standard
axioms propext, Classical.choice and Quot.sound; no sorry,
native_decide, or new axiom.
Triage
Tier 1 external named conjecture, preregistered in issue #10534 before the
probe. theorem; resolution proved. The public theorem has
proof_shape: content: the classification of step 2, the invariance of step
3 and the additivity of step 4 are new propositions on the live proof path.
Its admission basis is open-problem-resolution; utility none.
ASSUMED-UNVERIFIED
The reading of the entropy, of LC equivalence and of the four-star follows the paper; the bounded literature check does not establish exhaustive worldwide novelty, priority, or the absence of an independent proof.