On Theorems of Sinajova, Rankin and Kuperberg Concerning Spherical Point Configurations

A. Y. Alfakih

REVIEW 1 major objections 4 minor 16 references

On Theorems of Sinajova, Rankin and Kuperberg Concerning Spherical Point Configurations

T0 review · 1 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Four known theorems about spherical point configurations are reproved as corollaries of a single eigenvalue-multiplicity lemma for spherical Euclidean distance matrices.

desk verdict A neat, honest unification of three known theorems via EDM plus Perron-Frobenius; the Sinajova and Rankin proofs work, but the Kuperberg proof has a real gap in its block decomposition that looks repairable. read the letter →

arxiv 1908.00881 v2 pith:WFD6MPDI submitted 2019-08-02 math.MG

A. Y. Alfakih This is my paper · ORCID

classification math.MG MSC 15A1851K0552C17

keywords orthogonalrepresentationofgraphsdistancegeometryspherepackingdispersionproblemEuclideanmatricesPerron-Frobeniustheoremsphericalpointconfigurations

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This note claims that four classical theorems about point configurations on a unit sphere—a graph orthonormal-representation dimension theorem, two packing bounds of Rankin, and Kuperberg's splitting theorem—are all consequences of a single linear-algebraic fact about spherical Euclidean distance matrices. The fact, proved as Lemma 3.1, is that for any unit spherical EDM D written as D = 2(E - I) + 2Δ, the normalized matrix Δ has largest eigenvalue 1, and the embedding dimension of the configuration equals n minus the multiplicity of that eigenvalue. From this, the four theorems follow by applying the Perron-Frobenius theorem to nonnegative irreducible blocks of Δ. A sympathetic reader should care because the paper replaces several separate geometric arguments with one linear algebra calculation, making the common structure of the theorems visible.

What carries the argument

The load-bearing object is the decomposition of a unit spherical EDM D as D = 2(E - I) + 2Δ, where E is the all-ones matrix and Δ is a symmetric nonnegative matrix with zero diagonal. Lemma 3.1 is the identity λmax(Δ) = 1, with the vector w satisfying Dw = e serving as an eigenvector for λmax(Δ), and the rank formula r = n - m(λmax(Δ)). Because the Gram matrix is B = E - D/2 = I - Δ, this decomposition converts spherical geometry into nonnegative-matrix spectral data, so Perron-Frobenius controls irreducibility, positivity, and block structure.

What would settle it

Take any unit spherical configuration satisfying Kuperberg's hypotheses—say n points on the unit sphere in Rr with 2 ≤ n - r ≤ r and all pairwise squared distances at least 2—form D and Δ = D/2 + I - E, and compute the Perron-Frobenius block decomposition of Δ. If any irreducible block has spectral radius strictly less than 1 while λmax(Δ) = 1 has multiplicity n - r, the proof's equation (10) is unjustified; finding even one such example would invalidate the proof as written, while checking a regular crosspolytope should produce exactly r blocks of spectral radius 1.

Watch

Extended reading notes

Core claim

The central discovery is that the geometry of widely separated points on a unit sphere is encoded in the spectrum of a normalized nonnegative matrix built from the squared distances. Specifically, for any unit spherical Euclidean distance matrix D, setting D = 2(E - I) + 2Δ gives a symmetric nonnegative matrix Δ with zero diagonal; the Gram matrix of the configuration is I - Δ, so Δ has largest eigenvalue exactly 1, and the configuration's embedding dimension is n minus the multiplicity of λmax(Δ). The paper then shows that rank considerations force the contradiction in Rankin's n = r + 2 theorem, force the block decomposition in Kuperberg's splitting theorem, and, through the adjacency matrix of a graph, give the dimension formula d(G) = n - k for orthonormal representations. The theorems themselves are not new; the contribution is that each is presented as a corollary of the same eigenvalue-multiplicity identity combined with Perron-Frobenius theory.

Load-bearing premise

The proof of Kuperberg's theorem assumes that when λmax(Δ) = 1 has multiplicity n - r, the Perron-Frobenius block decomposition of Δ consists exactly of n - r irreducible blocks, each with spectral radius 1; if a block with spectral radius below 1 appears, the claimed orthogonal splitting does not follow from the argument as written.

Editorial extensions

If this is right

Rankin's n = r + 2 theorem follows immediately: if all pairwise squared distances exceeded 2, then Δ would be entrywise positive and hence irreducible, forcing the embedding dimension to be n - 1 instead of n - 2.
Kuperberg's splitting theorem follows from the block structure of Δ: the n - r irreducible blocks of spectral radius 1 correspond to mutually orthogonal subspaces L1,...,L_{n-r}, each carrying ri + 1 of the points.
Rankin's n = 2r theorem is the special case of Kuperberg's theorem where every block has size 2, forcing each pair to be antipodal and the configuration to be a regular r-crosspolytope.
The dimension formula d(G) = n - k for orthonormal graph representations is read off as m(λmax(Δ)) = k for the normalized adjacency matrix of the graph.
All four proofs share the same template: write D = 2(E - I) + 2Δ and apply Lemma 3.1, which unifies previously separate geometric arguments into one linear-algebraic scheme.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The eigenvalue-multiplicity formula suggests a general necessary condition for any spherical point configuration with prescribed minimal distance: the normalized distance matrix's largest eigenvalue 1 and its multiplicity constrain the number of points that can sit on a given sphere. This is an extension beyond the four theorems treated in the paper.
The proof's distinction between the origin lying in the relative interior versus relative boundary of a block's point set could be developed into a classification of the rigid-motion types admitted by Kuperberg's splitting, since it records whether the zero eigenvector block has positive support.
A likely testable extension is to replace the unit sphere by a sphere of radius ρ: the scaling of Δ should shift the eigenvalue condition, possibly yielding analogous packing statements for other minimal-distance thresholds. The paper does not pursue this.
For orthonormal graph representations, the same machinery might generate dimension bounds for graphs with weighted edges, since edge weights enter Δ and Perron-Frobenius still controls the spectral radius.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

First thing to know: this note gives short linear-algebraic proofs of three known theorems in spherical point configurations, using spherical Euclidean distance matrices and the Perron-Frobenius theorem. The results are not new; the value is the unified machinery. And it mostly works. The Sinajova proof is clean, the Rankin proof is solid, and the example at the end illustrates the geometry nicely. But the Kuperberg proof has a genuine gap.

The problem is equation (10). From the Perron-Frobenius theorem and the fact that λ_max(Δ) has multiplicity n−r ≥ 2, the author concludes Δ is block diagonal with exactly n−r irreducible blocks, each with spectral radius 1. That does not follow. Perron-Frobenius gives at least n−r such blocks; there can also be irreducible blocks with spectral radius less than 1. The stress-test example—three blocks, two with spectral radius 1 and one with spectral radius 1/2—satisfies all the hypotheses of Theorem 5.1 but has three irreducible components, not n−r = 2. So as written, the split into the subspaces L_i is not established. The theorem itself is known, and the gap looks repairable: you can likely merge the small block with an adjacent λ=1 block and recover the same conclusion. But that repair is not in the manuscript.

Other issues are minor. There is a typo in Section 3 (an inequality direction in the nonexistence part), and the author rightly acknowledges that Naszódi and Polyanskii already used the same EDM-plus-Perron-Frobenius idea for Rankin's theorem. So the novelty is the unified treatment and the Sinajova proof, which is nice but not a game-changer.

Who should read this? Anyone teaching distance geometry or spherical codes will like the Sinajova proof and the elegant Lemma 3.1. The Kuperberg gap makes the note a little risky as-is, but it is specific and fixable. I would send it to a refereed venue—a short note like this deserves a referee—with a clear request to fix the block decomposition argument. The rest of the proof is easy to verify.

Referee Report

1 major / 4 minor

Summary. This note proposes a unified linear-algebraic proof of four classical theorems on spherical point configurations: Sinajova's theorem on orthonormal representations of graphs, Rankin's two dispersion theorems for n=r+2 and n=2r, and Kuperberg's orthogonal-splitting theorem for r+2≤n≤2r. The common device is to write a unit spherical Euclidean distance matrix D as D=2(E−I)+2Δ. Lemma 3.1 shows that under this normalization λmax(Δ)=1 and the embedding dimension is n minus the multiplicity of that eigenvalue. The Perron–Frobenius theorem is then applied to the nonnegative matrix Δ. Section 3 proves Sinajova's theorem, Section 4 proves Rankin's n=r+2 theorem, and Section 5 derives Kuperberg's theorem, with Rankin's n=2r theorem presented as a special case.

Significance. If all proofs were correct, the paper would be a compact and attractive unification: the EDM formulation makes the role of the Perron–Frobenius theorem transparent, and the proofs are genuinely elementary and free of fitted parameters. The derivations of Sinajova's theorem and Rankin's n=r+2 theorem are clean and sound. However, the proof of Kuperberg's theorem currently rests on an unjustified block-decomposition claim in Section 5, and since Rankin's n=2r theorem is deduced from Kuperberg's theorem, the paper's central claim is not fully established as written. The gap appears repairable, but the repair is not present in the manuscript.

major comments (1)

[Section 5, Eq. (10)] The Perron–Frobenius theorem does not imply that QΔQᵀ has exactly n−r nonzero irreducible diagonal blocks all with λmax=1. It implies only that there are at least n−r irreducible components with Perron root 1; additional irreducible components with Perron root strictly less than 1 are allowed. Such a case satisfies every hypothesis of Theorem 5.1: take n=6 and Δ=diag(A,A,C), where A=[[0,1],[1,0]] and C=[[0,1/2],[1/2,0]]. Then λmax(Δ)=1 with multiplicity 2, so B=I−Δ has rank 4, and D=2(E−I)+2Δ is a unit spherical EDM of embedding dimension r=4; all off-diagonal entries of D are at least 2, and 2≤n−r≤r. Yet Δ has three nonzero irreducible components rather than the n−r=2 asserted in Eq. (10). Therefore the displayed decomposition is false in general, and the conclusion that the configuration splits into n−r simplex classes does not follow from the proof as written. Since Theorem 5.1 is also used to obtain Rankin's n=2r theorem, the proof of that result is likewise incomplete. A repair would need to handle the low-Perron-root blocks explicitly, for example by merging them into neighboring Perron components, but no such argument appears in the manuscript.

minor comments (4)

[Section 3, final paragraph] The displayed inequality "r ≤ n−k" is backwards; the reasoning in the preceding sentence gives m(λmax(Δ)) ≤ k, and hence r = n−m(λmax(Δ)) ≥ n−k, which is the lower bound needed for the theorem.
[Section 5, paragraph after Theorem 5.1] The phrase "at least one of the off-diagonal diagonal entries of D is 2" should be "at least one of the off-diagonal entries of D is 2".
[Example 5.1] "subsapces" is a typo for "subspaces".
[Lemma 5.1] The proof, described only as similar to Lemma 4.1, would be clearer with an explicit rank computation showing how the padded zero block contributes no additional rank; the statement is true under the standing assumption that D is a unit spherical EDM, so λmax of the padded matrix is 1.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the proofs use standard Euclidean distance matrix characterizations and Perron-Frobenius theory; the target theorems are external results being reproved, not assumed inputs.

full rationale

The paper's derivation chain is self-contained relative to standard EDM theory. Lemma 3.1 derives the eigenvalue condition lambda_max(Delta)=1 and the embedding dimension formula n - m(lambda_max(Delta)) directly from the unit spherical EDM definition via Theorems 2.1 and 2.2, using only that Dw=e and e^T w=1/2. No parameter is fitted to a subset of data and then renamed a prediction; there are no fitted parameters at all. The proofs of Sinajova's, Rankin's, and Kuperberg's theorems invoke only the Perron-Frobenius theorem and the block structure of nonnegative matrices, not the theorems being proved. The cited prior results are classical external theorems or standard EDM characterizations; the only nearby self-citation, Alfakih's book [1], is background material and is not load-bearing for the arguments. The lack of a stated proof of an assertion in the Kuperberg section about the block decomposition is a potential mathematical gap, but it is a correctness issue, not a circularity: the asserted conclusion is not assumed as an input anywhere. Thus there is no step in which an output is equivalent by construction to an input, no fitted input called a prediction, and no load-bearing self-citation chain.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No free parameters are introduced; the paper uses fixed constants from the standard EDM theory and names no new entities. The axioms are all standard background results from the cited literature.

assumptions (4)

standard math Characterization of Euclidean distance matrices (Theorem 2.1)
Cited to Schoenberg, Young-Householder, Critchley, and Gower; used to ensure D is an EDM when the Gram matrix is PSD.
standard math Characterization of spherical EDMs and radius formula (Theorem 2.2)
Cited to Gower; used to show constructed EDMs are unit spherical with radius 1.
standard math Perron-Frobenius theorem for nonnegative irreducible matrices
Used to assert simplicity and positivity of the Perron root and to infer reducibility when multiplicity exceeds 1.
standard math Existence of w with Dw = e for nonzero EDMs
Used repeatedly, e.g., Eq (3) and Lemma 3.1.

how reviews work

0 comments

Cite this review

Pith. "Pith review of On Theorems of Sinajova, Rankin and Kuperberg Concerning Spherical Point Configurations." pith.science (2026). https://pith.science/paper/WFD6MPDI

@misc{pith2026190800881,
  author       = {Pith},
  title        = {Pith review of: On Theorems of Sinajova, Rankin and Kuperberg Concerning Spherical Point Configurations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WFD6MPDI}},
  note         = {Machine review of arXiv:1908.00881}
}

read the original abstract

This note presents simple linear algebraic proofs of theorems due to Sinajova, Rankin and Kuperberg concerning spherical point configurations. The common ingredient in these proofs is the use of spherical Euclidean distance matrices and the Perron-Frobenius theorem.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages

[1]

A. Y. Alfakih. Euclidean distance matrices and their applications in rigi d- ity theory . Springer, 2018

work page 2018
[2]

Critchley

F. Critchley. On certain linear mappings between inner-product a nd squared distance matrices. Linear Algebra Appl. , 105:91–107, 1988

work page 1988
[3]

Davenport and G

H. Davenport and G. Haj´ os. Problem 35 (in Hungarian), 1951. M at. Lapok 2, 68

work page 1951
[4]

J. C. Gower. Euclidean distance geometry. Math. Sci. , 7:1–14, 1982

work page 1982
[5]

J. C. Gower. Properties of Euclidean and non-Euclidean distance matrices. Linear Algebra Appl. , 67:81–97, 1985

work page 1985
[6]

Kuperberg

W. Kuperberg. Optimal arrangements in packing congruent balls in a spherical container. Disc. Comput. Geom. , 37:205–212, 2007

work page 2007
[7]

Lov´ asz

L. Lov´ asz. On the shannon capacity of graphs. IEEE Trans. Infom. Theory, 25:1–7, 1979

work page 1979
[8]

Lov´ asz.Graphs and Geometry

L. Lov´ asz.Graphs and Geometry . AMS, Colloquium publications vol 65, 2019

work page 2019

Show all 16 references

[9]

O. R. Musin. Graphs and spherical two-distance sets. Euro. J. Comb. , 80:311-325, 2019

2019
[10]

Nasz´ odi and A

M. Nasz´ odi and A. Polyanskii. Perron and Frobenius meet Cara th´ eodory. arXiv 1901.00540v1 , 2019

1901 arXiv
[11]

T. D. Parsons and T. Pisanski. Vector representations of gra phs. Disc. Math., 78:143–154, 1989

1989
[12]

R. A. Rankin. The closest packing of spherical caps in n dimensions. Proc. Glasgow Math. Assoc. , 2:139–144, 1955

1955
[13]

I. J. Schoenberg. Remarks to Maurice Fr´ echet’s article: Sur la d´ eﬁnition axiomatique d’une classe d’espaces vectoriels distanci´ es applicables vec- toriellement sur l’espace de Hilbert. Ann. Math. , 36:724–732, 1935. 12

1935
[14]

ˇSiˇ najov´ a

E. ˇSiˇ najov´ a. A note on vector representation of graphs. Disc. Math. , 89:315–317, 1991

1991
[15]

Y. Wang, J. Yang, and H. Xu. On the connection between maximin distance designs and orthogonal designs. Biometrika, 105:471–477, 2018

2018
[16]

Young and A

G. Young and A. S. Householder. Discussion of a set of points in t erms of their mutual distances. Psychometrika, 3:19–22, 1938. 13

1938

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

A. Y. Alfakih. Euclidean distance matrices and their applications in rigi d- ity theory . Springer, 2018

work page 2018

[2] [2]

Critchley

F. Critchley. On certain linear mappings between inner-product a nd squared distance matrices. Linear Algebra Appl. , 105:91–107, 1988

work page 1988

[3] [3]

Davenport and G

H. Davenport and G. Haj´ os. Problem 35 (in Hungarian), 1951. M at. Lapok 2, 68

work page 1951

[4] [4]

J. C. Gower. Euclidean distance geometry. Math. Sci. , 7:1–14, 1982

work page 1982

[5] [5]

J. C. Gower. Properties of Euclidean and non-Euclidean distance matrices. Linear Algebra Appl. , 67:81–97, 1985

work page 1985

[6] [6]

Kuperberg

W. Kuperberg. Optimal arrangements in packing congruent balls in a spherical container. Disc. Comput. Geom. , 37:205–212, 2007

work page 2007

[7] [7]

Lov´ asz

L. Lov´ asz. On the shannon capacity of graphs. IEEE Trans. Infom. Theory, 25:1–7, 1979

work page 1979

[8] [8]

Lov´ asz.Graphs and Geometry

L. Lov´ asz.Graphs and Geometry . AMS, Colloquium publications vol 65, 2019

work page 2019