Perron and Frobenius meet Carath\'{e}odory
M\'arton Nasz\'odi, Alexandr Polyanskii

TL;DR
This paper introduces a novel method leveraging the Perron-Frobenius Theorem to prove Carathéodory-type theorems, potentially opening new avenues for related geometric results.
Contribution
It presents a new approach using matrix theory to prove classical geometric theorems, offering a fresh perspective and potential extensions.
Findings
New proof technique for Carathéodory-type theorems
Potential extension to Colourful Carathéodory Theorem
Open question on broader applicability
Abstract
We present a new approach of proving certain Carath\'{e}odory-type theorems using the Perron-Frobenius Theorem, a classical result in matrix theory describing the largest eigenvalue of a matrix with positive entries. One of the problems left open in this note is whether our approach may be extended to prove similar results in the area, in particular the Colourful Carath\'{e}odory Theorem.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Perron and Frobenius Meet Carathéodory
Márton Naszódi, Alexandr Polyanskii
Márton Naszódi,
iiiDCG EPFL; Inst. of Math., ELTE
Alexandr Polyanskii,
Abstract.
We present a new approach of proving certain Carathéodory-type theorems using the Perron–Frobenius Theorem, a classical result in matrix theory describing the largest eigenvalue of a matrix with positive entries.
One of the problems left open in this note is whether our approach may be extended to prove similar results in the area, in particular the Colourful Carathéodory Theorem.
1. Introduction
Carathéodory’s Theorem [Car07] is one of conerstones of combinatorial geometry as Eckhoff [Eck93] called it. The theorem claims that a point in the convex hull of a set is in the convex hull of at most points of ; see the definition of convex hull in Section 2. The survey [Eck93] contains a number of generalizations of the theorem. As it usually happens with fundamental statements, Carathéodory’s Theorem is closely connected with many other classical results in convex geometry such as Helly’s and Radon’s Theorems [Hel23, Rad21]. In this note, we provide a new proof of the theorem that shows a connections with a principal theorem in matrix theory, the Perron–Frobenius Theorem.
Our proofs use induction on the number of points (but not on the dimension). Two, most likely related, questions remain open. First, can induction be avoided in the proofs? Second, does our method yield a proof of other Carathéodory-type theorems, specifically, Bárány’s Colourful Carathéodory Theorem [Bár82].
The paper is organized as follows. In Section 2, we introduce notations and the tools that we use. In Section 3, we illustrate the key idea applying it to prove Rankin’s Theorem [Ran55, Theorem 1(iii, iv)]. In Section 4, we use the idea to prove Carathéodory’s and Steinitz’s Theorems [Car07, Ste13].
2. Preliminaries
2.1. Notations
We write , for a positive integer . The convex hull of a finite set of points , written , is the set
[TABLE]
A point of a set is an interior point of , if the set contains some open ball with center in .
The spectral radius of a square matrix , written , is the largest absolute value of its eigenvalues.
2.2. Tools
Since we do not need the most general form of the Perron–Frobenius Theorem [Per07, Fro12], we state only two of its corollaries.
Lemma 1** (Perron’s Theorem).**
For an -by- matrix with positive entries, the spectral radius is an eigenvalue of multiplicity , such that its eigenvector has positive entries.
Lemma 2** (Frobenius’s Theorem).**
For an -by- matrix with non-negative entries, the spectral radius is an eigenvalue such that one of its eigenvectors has non-negative entries.
Also, we need two finite-dimensional versions of the Hahn–Banach Theorem in Section 4.
Theorem 3**.**
If the origin of does not lie in the convex hull of points , then there exists a vector such that for all .
Theorem 4**.**
If the origin of is not an interior point of the convex hull of points , then there exists a vector such that for all .
3. Proof of Rankin’s Theorem
We state the two parts of Rankin’s Theorem as separate theorems.
Theorem 5** (Rankin).**
If is a set of non-zero vectors in such that the angle between any two of them is larger than , then .
Proof.
Suppose to the contrary that . Let be the Gram matrix of the vectors , that is, . Choose a positive such that for all , and set . The hypothesis of the theorem implies that all entries of are positive. By Lemma 1, the spectral radius is the largest eigenvalue of of multiplicity one.
Obviously, because is the Gram matrix of -dimensional vectors. Hence [math] is an eigenvalue of of multiplicity at least two, that is, is an eigenvalue of of multiplicity at least two. Since the Gram matrix is positive semidefinite, and thus, all its eigenvalues are non-negative, must be the largest eigenvalue of . Therefore, , contradicting the fact that the multiplicity of the largest eigenvalue is one. ∎
Theorem 6** (Rankin).**
If is a set of non-zero vectors in such that the angle between any two of them is at least , then .
Proof.
Suppose to the contrary that . Without loss of generality we can assume that the vectors are of unit length. Let be the Gram matrix of the vectors . Set . The hypothesis of the theorem implies that all entries of are non-negative. By Lemma 2, the spectral radius is the largest eigenvalue of .
Obviously, because is the Gram matrix of -dimensional vectors. Hence, [math] is an eigenvalue of of multiplicity at least , and thus, is an eigenvalue of of multiplicity at least . Since the Gram matrix is positive semidefinite, that is, [math] is its smallest eigenvalue, is the largest eigenvalue of . It means that . Let be the eigenvalues of , indexed in non-decreasing order. Thus we have
[TABLE]
However, the last inequality contradicts for all . ∎
4. Proofs of Caratheodory’s and Steinitz’s Theorems
Theorem 7** (Carathéodory’s Theorem).**
If the origin lies in the convex hull of points , then there is a set of size at most such that lies in the convex hull of .
Proof.
We use induction on . The base case, , being trivial, it is sufficient to show that if and for all , then .
By Theorem 3, for all , there is a vector such that for all . Choose a positive such that for all and set , where and . By Lemma 1, the spectral radius is the largest eigenvalue of of multiplicity one.
Obviously, because and are -by- matrices. Hence, [math] is an eigenvalue of of multiplicity at least two, that is, is an eigenvalue of of multiplicity at least two. It follows that , so , because the spectral radius cannot be less than a positive eigenvalue. Consider the eigenvector of the eigenvalue . By Lemma 1, its entries are positive. Therefore, we obtain
[TABLE]
Hence, is a -dimensional vector such that for all , because all entries of the vector are positive. So the points lie in an open half-space bounded by a hyperplane passing through the origin. Thus, their convex hull does not contain the origin. ∎
Theorem 8** (Steinitz’s Theorem).**
If the origin of is an interior point of the convex hull of points , then there is a set of size at most such that the point is interior of convex hull of .
Proof.
Again, we use induction on . The base case, , being trivial, it is sufficient to show that if and, for all , the origin is not an interior point of , then is not an interior point of .
By Theorem 4, for all , there is a non-zero vector such that for all . If for some , then the origin is not an interior point of the convex hull of . So, without loss of generality, we may assume that for all . Set , where and . By Lemma 2, the spectral radius is an eigenvalue of .
Obviously, because and are -by- matrices, and hence, [math] is an eigenvalue of of multiplicity at least , that is, is an eigenvalue of of multiplicity at least . Suppose that . Let be the eigenvalues of , indexed in non-decreasing order. Hence we have
[TABLE]
But the last inequality contradicts for all . Hence because the spectral radius cannot be less than a positive eigenvalue. Consider the eigenvector of the eigenvalue . By Lemma 2, its entries are non-negative entries. Therefore, we get
[TABLE]
Hence, for all . Moreover, is a non-zero vector, because among non-negative entries of there is at least one positive. So the points lie in a closed half-space bounded by a hyperplane passing through the origin, that is, the point is not interior of their convex hull. ∎
Appendix A Proofs of Separtion Theorems
Theorems 3 and 4 are fundamental in the study of convex sets. For completeness, we include their proofs.
A.1. Proof of Theorem 3
By compactness of the set , it has a point closest to the origin. Suppose that . Write . Thus we have
[TABLE]
Since the coefficient for in (2) is negative, for sufficiently small positive . Points and lie in , so , contradicting the choice of . Therefore, for all .
A.2. Proof of Theorem 4
Set . Denote the ray emanating from a point and passing through a point by .
We claim that there exists a ray such that . Otherwise, one can choose points such that the point is an interior point of .
Consider a sequence of points distinct from on the ray and converging to . By Theorem 3, there exists a sequence of unit vectors in such that for all and all positive integer . Since the unit sphere is compact, there exists a subsequence of unit vectors converging to a unit vector . Therefore, we obtain .
Acknowledgments
The work was done while the second named author was an academic visitor at EPFL.
The authors are grateful to János Pach and Imre Bárány for fruitful discussions.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[Bár 82] Imre Bárány, A generalization of Carathéodory’s theorem , Discrete Mathematics 40 (1982), no. 2-3, 141–152.
- 2[Car 07] Constantin Carathéodory, Über den Variabilitätsbereich der Koeffizienten von Potenzreihen, die gegebene Werte nicht annehmen , Mathematische Annalen 64 (1907), no. 1, 95–115.
- 3[Eck 93] Jürgen Eckhoff, Helly, Radon, and Carathéodory type theorems , Handbook of Convex Geometry, Part A, Elsevier, 1993, pp. 389–448.
- 4[Fro 12] Ferdinand Georg Frobenius, Über Matrizen aus nicht negativen Elementen , Sitzungsber. Königl. Preuss. Akad. Wiss. (1912), 456–477.
- 5[Hel 23] Eduard Helly, Über Mengen konvexer Körper mit gemeinschaftlichen Punkte , Jahresbericht der Deutschen Mathematiker-Vereinigung 32 (1923), 175–176.
- 6[Per 07] Oskar Perron, Zur Theorie der Matrices , Mathematische Annalen 64 (1907), no. 2, 248–263.
- 7[Rad 21] Johann Radon, Mengen konvexer Körper, die einen gemeinsamen Punkt enthalten , Mathematische Annalen 83 (1921), no. 1-2, 113–115.
- 8[Ran 55] Robert Alexander Rankin, The closest packing of spherical caps in n 𝑛 n dimensions , Glasgow Mathematical Journal 2 (1955), no. 3, 139–144.
