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.
Trace-moment comparison with Gaussian matrices
PDF p. 34: Proof of Proposition 6.3, trace comparison
Statement
Assume the Axiom of Choice for the adopted Gaussian-law construction. Let and be integers, and let . Let be the adjacency matrix of the loop-free Erdős–Rényi graph , let be an independent copy, and put and . Thus is symmetric, its diagonal is zero, and its entries above the diagonal are independent differences of independent Bernoulli variables. Let be the symmetric Gaussian comparator: its entries for are mutually independent standard real Gaussian variables, and for .
For put . Let consist of those ordered tuples with and with each unordered edge traversed an even number of times. For , let count its distinct unordered edges traversed exactly times, and put . Rotations and reversals are not identified unless they are equal as tuples. Write .
Then for every , and
Moreover, . Consequently .
Facts & Assumptions
The even-walk definition fixes the ordered tuples, their loop-free edges and traversal counts (Even closed walks and traversal multiplicities).
The adjacency difference has zero diagonal and off-diagonal positive odd moments zero and positive even moments (Symmetrization and moments of the adjacency difference). The differences for distinct unordered edges are mutually independent (Finite Bernoulli difference laws and parity of moments).
The trace of an even power expands over ordered closed walks. For a symmetric matrix with independent triangular entries and finite moments, the expectation of a walk product factors over its distinct unordered endpoint pairs (Trace powers, closed walks and independent edge moments).
Under Choice, the comparator has mutually independent standard Gaussian triangular entries (Gaussian comparison matrix in the moment argument, The Axiom of Choice). Their odd moments vanish, their positive even moments are at least one, and all their moments exist absolutely (Gaussian entry moments and triangular independence).
Proof
Given: The matrices, parameters and Choice assumption in the Statement.
Goal: Establish the walk expansion, corrected bounds and Gaussian comparison.
By [F3], expand the trace of over ordered closed walks and factor each expected walk product by unordered endpoint pair. A walk containing a diagonal step has zero product by [F2]. Among the remaining walks, an odd traversal multiplicity gives a zero factor by [F2]. For a surviving walk every distinct traversed edge contributes , so its contribution is . These surviving walks are exactly the tuples in from [F1], including when . Finite summation proves the displayed equality.
Each has steps and at least one traversed edge. Every traversed edge has positive even multiplicity, so . Also , since . Thus ; these inequalities include because all exponents are positive. Summing and using step 1.1 gives both bounds.
Apply [F3] to . By [F4], every walk with an odd multiplicity has zero expected product, while every walk with all multiplicities even has a product of moments at least one. Diagonal steps are allowed in this Gaussian expansion, so its nonnegative surviving terms include every loop-free even tuple in . Therefore . Multiplying by and using step 2.1 gives the final comparison. The stated expansion and comparisons follow.
Source notes
The article attributes the expansion and comparison to the passage cited above. Its displayed upper bound with is reversed term by term: and . The corrected upper bound here uses ; it does not imply the article's subsequent norm estimate with the same power of after taking a -th root.
Depends on
- Even closed walks and traversal multiplicities
- Symmetrization and moments of the adjacency difference
- Gaussian comparison matrix in the moment argument
- Trace powers, closed walks and independent edge moments
- Gaussian entry moments and triangular independence
- Finite Bernoulli difference laws and parity of moments
- The Axiom of Choice
- The Erdős-Rényi finite random graph G(n,p)
- The adjacency matrix of a finite simple graph
Dependency tree · 2 levels
Each result is drawn at its shortest distance from this item. Arrows run from prerequisites to the items that use them; this item has a heavier outline. External source nodes link directly to the cited text. The view shows up to two dependency steps; open a prerequisite to continue.
Sources
- Expander graphs are globally synchronizing · PDF p. 34, proof of Proposition 6.3, trace comparison