← All problems

Erdős Problem #991 — Discrepancy of maximum-product points on the sphere

62 years old · traced to

Let S2={x∈R3:∣x∣=1}S^2=\{x\in\mathbb R^3:|x|=1\}, with Euclidean distance, and for each integer n≥1n\ge 1 let An={w1,…,wn}⊂S2A_n=\{w_1,\ldots,w_n\}\subset S^2 be any set maximizing the product of pairwise distances among all nn-point subsets of S2S^2:

∏1≤i<j≤n∣wi−wj∣.\prod_{1\le i<j\le n}|w_i-w_j|.

For a spherical cap CC, let αC\alpha_C denote its surface area divided by the total surface area of S2S^2. Is it true that, for every sequence of such maximizing sets (An)n≥1(A_n)_{n\ge1},

max⁡C∣ ∣An∩C∣−nαC∣=o(n)(n→∞),\max_C\left|\,|A_n\cap C|-n\alpha_C\right|=o(n)\qquad(n\to\infty),

where the maximum is over all spherical caps? Erdős (1964) Erdős Problems #991

Equivalently, these sets minimize the logarithmic energy E0(An)=∑i≠jlog⁡(1/∣wi−wj∣)E_0(A_n)=\sum_{i\ne j}\log(1/|w_i-w_j|), since E0(An)=−2log⁡∏i<j∣wi−wj∣E_0(A_n)=-2\log\prod_{i<j}|w_i-w_j|. The question asks whether the unnormalized cap-counting error is sublinear in the number of points. Marzo and Mas, Discrepancy of Minimal Riesz Energy Points

Background

Erdős raised the question in 1964 after introducing the spherical-cap discrepancy of a finite point set: the largest difference between the number of points in a cap and the number predicted by uniform surface measure. He observed that the least possible discrepancy over all configurations should grow with the number of points, and separately asked whether the highly structured configurations maximizing the product of pairwise distances nevertheless have discrepancy o(n)o(n). The latter, rather than the minimum-discrepancy problem, is the question here. Erdős (1964)

The product maximization is the logarithmic-energy case of a broader family of discrete energy-minimization problems. For logarithmic and Riesz energy on the sphere, the minimizing point sets are classically known to become uniformly distributed as their size tends to infinity; equivalently, their normalized spherical-cap discrepancy tends to zero. That qualitative convergence gives an error of o(n)o(n) in the unnormalized count, while quantitative discrepancy estimates give a rate. A useful mechanism is to control a Sobolev discrepancy associated with the energy and then test it against functions approximating cap indicators. Marzo and Mas, Discrepancy of Minimal Riesz Energy Points

References

P. Erdős, Problems and results on diophantine approximations, Compositio Mathematica 16 (1964), 52–65.

Source: P. Erdős, Problems and results on diophantine approximations, Compositio Mathematica 16 (1964), 52–65.

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.