Maximizing expected powers of the angle between pairs of points in projective space

DOI10.1007/S00440-022-01108-1arXiv2007.13052MaRDI QIDQ6345837FDOQ6345837

Authors: Tongseok Lim, Robert J. McCann 

Publication date: 26 July 2020

Abstract: Among probability measures on $d$-dimensional real projective space, one which maximizes the expected angle $arccos(fracx|x|cdotfracy|y|)$ 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 T'oth cite{FT59}. If true, this conjecture evidently implies the same measure maximizes the expectation of $arcco{s}^{a}lpha(fracx|x|cdotfracy|y|)$ for any exponent $alpha>1$. The kernel $arcco{s}^{a}lpha(fracx|x|cdotfracy|y|)$ represents the objective of an infinite-dimensional quadratic program. We verify discrete and continuous versions of this {milder} conjecture in a non-empty range $alpha>alph{a}_{Delt{a}^d}ge1$, and establish uniqueness of the resulting maximizer $hatmu$ up to rotation. We show $hatmu$ no longer maximizes when $alpha<alph{a}_{Delt{a}^d}$. At the endpoint $alpha=alph{a}_{Delt{a}^d}$ of this range, we show another maximizer $mu$ must also exist which is not a rotation of $hatmu$. For the continuous version of the conjecture, an appendix provided by Bilyk et al in response to an earlier draft of this work combines with the present improvements to yield $alph{a}_{Delt{a}^d}<2$. The original conjecture $ald=1$ remains open (unless $d=1$). However, in the maximum possible range $alpha>1$, we show $hatmu$ and its rotations maximize the aforementioned expectation uniquely on a sufficiently small ball in the $L^{i}nfty$-Kantorovich-Rubinstein-Wasserstein metric $d_{i}nfty$ from optimal transportation; the same is true for any measure $mu$ which is mutually absolutely continuous with respect to $hatmu$, but the size of the ball depends on {$alpha,d$, and} $|fracdhatmudmu{|}_{infty}$.

Mathematics Subject Classification ID

Optimal transportation (49Q22) Nonconvex programming, global optimization (90C26) Other designs, configurations (05B30) Inequalities and extremum problems involving convexity in convex geometry (52A40) Variational methods for problems in mechanics (70G75) Variational inequalities (global problems) in infinite-dimensional spaces (58E35)

This page was built for publication: Maximizing expected powers of the angle between pairs of points in projective space

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6345837)