RemarkRemark: AI-adaptedProof: Literature-sourced

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.

Scoped independent low-rank landscape reports

Statement

For integers $n \geq 1$ and $r \geq 2$ , a sphere configuration consists of $n$ unit vectors in $R^{r}$ ; their Gram matrix has entries given by their pairwise inner products. A Burer–Monteiro factorization represents a positive semidefinite matrix with unit diagonal and rank at most $r$ by such a Gram matrix $Y Y^{T}$ , with $Y$ having $r$ columns. Here a landscape means the behavior of an objective on its feasible configurations, including its local extrema and critical points. On the feasible smooth manifold $M_{n, r}$ of unit-row matrices, let $f : M_{n, r} \to R$ be a smooth objective. A second-order critical point is a feasible $Y$ with $d f_{Y} = 0$ and ${Hess}_{Y} (f) (V, V) \geq 0$ for every $V \in T_{Y} M_{n, r}$ . Here ${Hess}_{Y} (f)$ is the intrinsic Hessian at a critical point, defined by the second derivatives in any smooth chart; the second variation in direction $V$ means its quadratic value ${Hess}_{Y} (f) (V, V)$ . A global minimizer has objective value no larger than every feasible point.

McRae, Abdalla, Bandeira and Boumal report near-threshold asymptotic guarantees for three specified binary synchronization and clustering models with sufficiently large finite rank. Rakoto Endor and Waldspurger report a Laplacian-conditioning criterion for global optimality of every second-order critical point in their low-rank semidefinite-program setting. McRae subsequently reports results for every $r \geq 2$ in several specified asymptotic recovery models and an extension of the conditioning criterion using diagonal preconditioning, together with results for random signed oscillator networks.

Statement references

Sphere configurations and tangent variations

Signed vector-spin objectives and low-rank Gram factors

The intrinsic Hessian of a smooth function at a critical point

External input

The three reports above are imported only at the level of their authors' abstracts. Their exact inequalities, model parameters and proofs are not supplied here, and these reports alone are not an unqualified circle-rank exact-recovery theorem or a theorem applicable to arbitrary graphs. An application requires the relevant full theorem and verification of its hypotheses.

Depends on

Used by · 1

Reported progress on benign landscapes and an alternative synchronization proof

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.

External dependencies

Nonconvex landscapes for Z2 synchronization and graph clustering are benign near exact recovery thresholds · Abstract

For three specified random models, the authors report asymptotic benign-landscape guarantees arbitrarily close to the previously established exact-recovery threshold with sufficiently large finite rank; they also report deterministic conditions and robustness to monotone adversaries.

Hypotheses: The abstract concerns the stated Z2 synchronization and graph-clustering models, their noise and graph assumptions, and a sufficiently large finite rank parameter; exact inequalities are not reproduced here.

Reading limits: The abstract was read. The theorem statements, constants, proofs and adversarial assumptions were not checked.

Benign landscape for Burer-Monteiro factorizations of MaxCut-type semidefinite programs · Abstract

For MaxCut-type semidefinite programs admitting a low-rank solution, the authors report a sharp condition on the associated Laplacian's conditioning under which every second-order critical point of the factorized problem is globally minimizing.

Hypotheses: The source's low-rank MaxCut-type semidefinite program and its Laplacian conditioning condition; the precise inequality is not reproduced here.

Reading limits: The version 2 abstract was read. The precise theorem hypotheses and proof were not checked.

Benign landscapes for synchronization on spheres via normalized Laplacian matrices · Abstract

McRae reports exact recovery at every second-order critical point for each rank r at least two in information-theoretically feasible asymptotic regimes of several specified models, as well as optimal asymptotic results for random signed Erdős–Rényi oscillator networks; the reported deterministic condition extends Rakoto Endor and Waldspurger's condition using diagonal preconditioning.

Hypotheses: The specified spherical least-squares models and asymptotic regimes, respectively the specified random signed oscillator networks; the exact model parameters and deterministic inequality are not reproduced here.

Reading limits: The version 2 abstract was read. Individual theorem hypotheses, constants, proofs and any identification with the article's Theorem 1.2 were not checked.