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.
Connected fiber graphs with negative semidefinite intersection form
Statement
Let be a finite connected loopless multigraph with vertices and edge multiplicities for . Define the symmetric matrix by and . Suppose is negative semidefinite, its kernel is one-dimensional, and it contains a vector with every coordinate strictly positive. Then is one of the affine Dynkin graphs (), (), , , or . Here is two vertices joined by two edges; for is a cycle of vertices; is the four-legged star; for has two trivalent vertices joined by a path of edges, with two terminal leaves at each end; and the three affine exceptional graphs have one trivalent vertex with arm lengths , , and , respectively. Primitive means that the greatest common divisor of the integer coordinates is . In each case the primitive positive integral kernel vector is unique.
Statement references
Facts & Assumptions
Given: , its edge multiplicities, the associated matrix , and the three matrix hypotheses in the Statement.
Negative semidefinite means for every ; a real symmetric matrix has an orthonormal eigenbasis, so for a negative-semidefinite matrix implies (Real spectral theorem: a self-adjoint endomorphism of a finite-dimensional real inner product space has an orthonormal eigenbasis).
Proof
Given: , its edge multiplicities, the associated matrix , and the three matrix hypotheses in the Statement.
[F1] Negative semidefinite means for every ; a real symmetric matrix has an orthonormal eigenbasis, so for a negative-semidefinite matrix implies (Real spectral theorem: a self-adjoint endomorphism of a finite-dimensional real inner product space has an orthonormal eigenbasis).
Proof technique: classify proper null subgraphs, then use the Schur complement at the unique trivalent vertex.
A proper induced subgraph cannot have a nonzero null vector. Its principal matrix is negative semidefinite by [F1]. Extending a null vector by zero gives a vector with , hence . Since is generated by a strictly positive vector, this extended vector cannot vanish at any vertex; therefore the subgraph was all of .
No edge has multiplicity at least three: on its two endpoints the principal matrix has positive eigenvalue when . If an edge has multiplicity two, is a nonzero null vector of its principal matrix, so [step 1.1] forces that two-vertex graph to be all of ; this is . Henceforth every edge is simple.
If a vertex has degree , the principal star matrix has positive quadratic value at the vector with center and each leaf : the value is , contradicting [F1]. If a vertex has degree four, the star vector with center and leaves is null, so [step 1.1] forces the entire graph to be that star, . Otherwise the maximum degree is at most three.
By steps 2.1 and 2.2 the graph is simple with maximum degree at most three. If it contains a cycle, choose a shortest simple cycle; it has no chord, since any chord would cut it into a shorter cycle. The vector equal to one on this induced cycle and zero elsewhere has quadratic value zero: each cycle vertex has exactly two selected incident edges and the cycle has no chords. By [step 1.1] its vertex set is all of , hence for its number of vertices . If is a tree with no vertex of degree three, it is a path; for a path on vertices, for every nonzero , so it has no kernel.
If the tree has at least two trivalent vertices, take two consecutive trivalent vertices and the path between them, together with one adjacent off-path vertex at each of the two unused branches at both ends. This subgraph is an affine graph. Its positive null vector has value on the path and the two trivalent vertices and value on the four leaves; the row equations give , and the edge-square identity shows it is negative semidefinite with one-dimensional kernel. Thus [step 1.1] forces , giving .
By steps 2.1–2.2, 3.1, and 4.1, the only remaining case is a tree with exactly one trivalent vertex. Let its three arm lengths, counted in vertices beyond the central vertex, be . For , each arm block is the positive definite path matrix with diagonal and adjacent entries . Solving the path equations gives the endpoint entry of its inverse as (the solution of is ). Eliminating the three arm blocks leaves central Schur complement . Since is positive semidefinite of corank one, this complement is zero, so .
If , each reciprocal is at most ; equality of their sum with forces , the affine graph. If , then . With , gives and gives ; would require , while makes the sum at most . These are exactly the affine and graphs. This exhausts the single-trivalent-vertex case.
The integer matrix has a one-dimensional real kernel, so its rational nullspace is also one-dimensional because rank is unchanged on extending scalars from to . It therefore has a nonzero rational kernel vector; clearing denominators and choosing its sign gives an integral vector proportional to the strictly positive vector in the hypothesis. The one-dimensional kernel makes this positive integral vector unique up to positive integer scale, and dividing its coordinates by their greatest common divisor gives the unique primitive positive integral kernel vector. The classification in steps 3.1–6.1 includes the double-edge , the four-legged , and all three exceptional arm triples.