Abstract
Among probability measures on d-dimensional real projective space, one which maximizes the expected angle \(\arccos (\frac{x}{|x|}\cdot \frac{y}{|y|})\) between independently drawn projective points x and y was conjectured to equidistribute its mass over the standard Euclidean basis \(\{e_0,e_1,\ldots , e_d\}\) by Fejes Tóth (Acta Math Acad Sci Hung 10:13–19, 1959. https://doi.org/10.1007/BF02063286). If true, this conjecture evidently implies the same measure maximizes the expectation of \(\arccos ^\alpha (\frac{x}{|x|}\cdot \frac{y}{|y|})\) for any exponent \(\alpha > 1\). The kernel \(\arccos ^\alpha (\frac{x}{|x|}\cdot \frac{y}{|y|})\) represents the objective of an infinite-dimensional quadratic program. We verify discrete and continuous versions of this milder conjecture in a non-empty range \(\alpha > \alpha _{\Delta ^d} \ge 1\), and establish uniqueness of the resulting maximizer \({\hat{\mu }}\) up to rotation. We show \({\hat{\mu }}\) no longer maximizes when \(\alpha <\alpha _{\Delta ^d}\). At the endpoint \(\alpha =\alpha _{\Delta ^d}\) of this range, we show another maximizer \(\mu \) must also exist which is not a rotation of \({\hat{\mu }}\). For the continuous version of the conjecture, an “Appendix A” provided by Bilyk et al in response to an earlier draft of this work combines with the present improvements to yield \(\alpha _{\Delta ^d}<2\). The original conjecture \({\alpha _{\Delta ^d}}=1\) remains open (unless \(d=1\)). However, in the maximum possible range \(\alpha >1\), we show \({\hat{\mu }}\) and its rotations maximize the aforementioned expectation uniquely on a sufficiently small ball in the \(L^\infty \)-Kantorovich–Rubinstein–Wasserstein metric \(d_\infty \) from optimal transportation; the same is true for any measure \(\mu \) which is mutually absolutely continuous with respect to \({\hat{\mu }}\), but the size of the ball depends on \(\alpha ,d\), and \(\Vert \frac{d {\hat{\mu }}}{d\mu }\Vert _{\infty }\).
Similar content being viewed by others
References
Aigner, M.: Turán’s graph theorem. Am. Math. Mon. 102, 808–816 (1995). https://doi.org/10.2307/2974509
Alexander, R., Stolarsky, K.B.: Extremal problems of distance geometry related to energy integrals. Trans. Amer. Math. Soc. 193, 1–31 (1974). https://doi.org/10.2307/1996898
Balagué, D., Carrillo, J.A., Laurent, T., Raoul, G.: Dimensionality of local minimizers of the interaction energy. Arch. Ration. Mech. Anal. 209, 1055–1088 (2013). https://doi.org/10.1007/s00205-013-0644-6
Benedetto, J.J., Fickus, M.: Finite normalized tight frames. Adv. Comput. Math. 18(2–4), 357–385 (2003). https://doi.org/10.1023/A:1021323312367
Bilyk, D., Dai, F.: Geodesic distance Riesz energy on the sphere. Trans. Am. Math. Soc. 372, 3141–3166 (2019). https://doi.org/10.1090/tran/7711
Bilyk, D., Dai, F., Matzke, R.: The Stolarsky principle and energy optimization on the sphere. Constr. Approx. 48, 31–60 (2018). https://doi.org/10.1007/s00365-017-9412-4
Bilyk, D., Glazyrin, A., Matzke, R., Park, J., Vlasiuk, O.: Energy on spheres and discreteness of minimizing measures. J. Funct. Anal. 280, 108995 (2021). https://doi.org/10.1016/j.jfa.2021.108995
Bilyk, D., Glazyrin, A., Matzke, R., Park, J., Vlasiuk, O.: Optimal measures for p-frame energies on the sphere (2019). arXiv:1908.00885
Bilyk, D., Matzke, R.W.: On the Fejes Tóth problem about the sum of angles between lines. Proc. Am. Math. Soc. 147, 51–59 (2019). https://doi.org/10.1090/proc/14263
Birkhoff, G.: Three observations on linear algebra. Univ. Nac. Tucumán. Revista A 5, 147–151 (1946)
Björck, G.: Distributions of positive mass, which maximize a certain generalized energy integral. Ark. Mat. 3, 255–269 (1956). https://doi.org/10.1007/BF02589412
Braides, A.: \(\Gamma \)-convergence for Beginners. Oxford Lecture Series in Mathematics and Its Applications, vol. 22. Oxford University Press, Oxford (2002). https://doi.org/10.1093/acprof:oso/9780198507840.001.0001
Carrillo, J.A., Figalli, A., Patacchini, F.S.: Geometry of minimizers for the interaction energy with mildly repulsive potentials. Ann. Inst. H. Poincaré Anal. Non Linéaire 34, 1299–1308 (2017). https://doi.org/10.1016/j.anihpc.2016.10.004
Fejes Tóth, L.: Über eine Punktverteilung auf der Kugel. Acta Math. Acad. Sci. Hung. 10, 13–19 (1959). https://doi.org/10.1007/BF02063286
Fodor, F., Vígh, V., Zarnócz, T.: On the angle sum of lines. Arch. Math. (Basel) 106, 91–100 (2016). https://doi.org/10.1007/s00013-015-0847-1
Kang, K., Kim, H.K., Seo, G.: Cardinality estimation of support of the global minimizer for the interaction energy with mildly repulsive potentials. Physica D 399, 51–57 (2019). https://doi.org/10.1016/j.physd.2019.04.004
Lim, T., McCann, R.J.: Isodiametry, variance and regular simplices from particle interactions. Arch. Ration. Mech. Anal. 241, 553–576 (2021). https://doi.org/10.1007/s00205-021-01632-9
Lim, T., McCann, R.J.: On Fejes Tóth’s conjectured maximizer for the sum of angles between lines. Appl. Math. Optim. 84(3), 3217–3227 (2021). https://doi.org/10.1007/s00245-020-09745-5
Lim, T., McCann, R.J.: On the cardinality of sets in \({\mathbf{R}}^d\) obeying a slightly obtuse angle bound. Under revision for SIAM J. Discrete Math
McCann, R.J.: Stable rotating binary stars and fluid in a tube. Houston J. Math. 32, 603–632 (2006)
Pólya, G., Szegö, G.: Über den transfiniten Durchmesser (Kapazitätskonstante) von ebenen und räumlichen Punktmengen. J. Reine Angew. Math. 165, 4–49 (1931). https://doi.org/10.1515/crll.1931.165.4
Turán, P.: Eine Extremalaufgabe aus der Graphentheorie (Hungarian. German Summary). Mat. Fiz. Lapok 48, 436–452 (1941)
Villani, C.: Topics in Optimal Transportation. Graduate Studies in Mathematics, vol. 58. American Mathematical Society, Providence (2003)
Vlasiuk, O.: Discreteness of the minimizers of weakly repulsive interaction energies on Riemannian manifolds (2020). arXiv:2003.01597
Acknowledgements
The authors are not aware of any conflicts of interest; they have no financial or proprietary interests in any material discussed in this article. TL is a faculty member in Purdue University’s Krannert School of Management. RJM is a Canada Research Chair at the University of Toronto. TL’s work was also supported in part by ShanghaiTech University, the University of Toronto and its Fields Institute for the Mathematical Sciences. RM acknowledges partial support of this research by the Canada Research Chairs Program and Natural Sciences and Engineering Research Council of Canada Grants RGPIN 2015-04383 and 2020-04162. Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
With an appendix by Dmitriy Bilyk, Alexey Glazyrin, Ryan Matzke, Josiah Park and Oleksandr Vlasiuk.
TL is grateful for the support of ShanghaiTech University, and in addition, to the University of Toronto and its Fields Institute for the Mathematical Sciences, where parts of this work were performed. RM acknowledges partial support of his research by the Canada Research Chairs Program and Natural Sciences and Engineering Research Council of Canada Grants RGPIN 2015-04383 and 2020-04162. ©2021 by the authors.
Appendix A. By Dmitriy Bilyk, Alexey Glazyrin, Ryan Matzke, Josiah Park, and Oleksandr Vlasiuk
Appendix A. By Dmitriy Bilyk, Alexey Glazyrin, Ryan Matzke, Josiah Park, and Oleksandr Vlasiuk
Theorem A.1
(Estimating the threshold exponent: \({\alpha _{\Delta ^d}}\le 2\)) For every \(\alpha \ge 2\), the set of maximizers of (1.5) is precisely \(\mathcal {P}_{{\Delta }}^=(\mathbf{S}^d)\).
Proof
For \(t \in [-1,1]\) set \(f_\alpha (t) = (\frac{2}{\pi } \arccos |t| )^\alpha \) and \(g(t)=1-t^2\). We claim
Indeed, setting \(h_\alpha (t) := g(t)^{1/\alpha } - f_\alpha (t)^{1/\alpha }\), the computation \(h_2''(t) = (2t\pi ^{-1} -1)(1-t^2)^{-3/2}\) shows \(h_2(t)\) to be strictly concave on the interval \(t \in [0,1]\) and to vanish at both endpoints. For \(\alpha = 2\) this establishes (A.1). For \(\alpha >2\), the same conclusion then follows from \(\frac{dh_\alpha }{d\alpha }(t)\ge 0\). For \(\alpha <2\) we have \(\lim _{t \nearrow 1} f_\alpha '(t)=- \infty <g'(1)\) and \(f_\alpha (1)=0=g(1)\), thus domination of \(f_\alpha \) by g fails, confirming the reverse implication in (A.1).
Now (1.4) may be rewritten in the form \(E_\alpha (\mu ) = {F_{f_\alpha }}(\mu )\) where
Let \(\mu \in \mathcal {P}(\mathbf{S}^d)\), \(\sigma \) be the uniform probability on \(\mathbf{S}^d\), and \({\hat{\mu }} \in \mathcal {P}_{{\Delta }}^=(\mathbf{S}^d)\). For \(\alpha \ge 2\) we claim
where the middle two (in)equalities for g are known and reproved below, while the first and last follow from (A.1)—which also makes the first inequality strict unless \(\mu \) lies in the narrow closure of \(\mathcal {P}_{{\Delta }}(\mathbf{S}^d)\). On the other hand, for \(\mu \in \overline{\mathcal {P}_{{\Delta }}(\mathbf{S}^d)}\), (1.7) implies that the second inequality in (A.2) becomes strict unless \(\mu \in \mathcal {P}_{{\Delta }}^=(\mathbf{S}^d)\).
It remains to establish the middle two (in)equalities in (A.2) for g, which can be done in various ways, c.f. [4]. For example, defining the symmetric \((d+1) \times (d+1)\) matrix \(I(\mu )\) by
the Cauchy–Schwartz inequality for the Hilbert-Schmidt norm yields
since on \(\mathbf{S}^d\). If \(I(\mu )\) is a multiple of the identity matrix—as for \(\mu \in \{\hat{\mu },\sigma \}\)—then equality holds. This establishes (A.2) and completes the proof. \(\square \)
Rights and permissions
About this article
Cite this article
Lim, T., McCann, R.J. Maximizing expected powers of the angle between pairs of points in projective space. Probab. Theory Relat. Fields 184, 1197–1214 (2022). https://doi.org/10.1007/s00440-022-01108-1
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00440-022-01108-1
Keywords
- Infinite-dimensional quadratic programming
- Optimization in curved spaces
- Interaction energy minimization
- Spherical designs
- Projective space
- Extremal problems of distance geometry
- Great circle distance
- Attractive-repulsive potentials
- Mild repulsion limit
- Riesz energy
- \(L^\infty \)-Kantorovich–Rubinstein–Wasserstein metric
- \(d_\infty \)-local
- Frame