Abstract:We compute the spherical cap discrepancy of the Diamond ensemble (a set of evenly distributed spherical points) as well as some other quantities. We also define an area regular partition on the sphere where each region contains exactly one point of the Diamond ensemble. For a concrete choice of parameters, we prove that the Diamond ensemble provides the best spherical cap discrepancy known until date for a deterministic family of points.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is about the uniformity of the distribution of point sets on the sphere. In particular, for a class of point sets called "Diamond ensemble", it calculates their spherical cap discrepancy. Spherical cap discrepancy is one of the indicators to measure the uniformity of the distribution of point sets on the sphere, which describes the deviation between the point set and the ideal uniform distribution. Specifically, the author aims to:
1. **Calculate the spherical cap discrepancy of Diamond ensemble**: The author not only focuses on how to calculate the spherical cap discrepancy of this specific point set, but also explores other related quantities.
2. **Define an area - regular partition**: The author proposes a method of spherical area - regular partition based on Diamond ensemble, ensuring that each region contains and only contains one point.
3. **Prove the superiority of Diamond ensemble**: For some specific parameter selections, the author proves that Diamond ensemble provides the best - known spherical cap discrepancy results so far, especially outstanding in deterministic point sets.
### Key Formulas
- **Definition of spherical cap discrepancy**:
\[
D_{\text{sup},\text{cap}}(\omega_N)=\sup_{C\in\text{cap}}\left|\frac{\#(\omega_N\cap C)}{N}-\frac{\mu(C)}{\mu(S^2)}\right|
\]
where $\omega_N$ is a point set on the sphere, $C$ is a set of spherical caps, $\mu(C)$ is the measure of the spherical cap $C$, and $\mu(S^2)$ is the measure of the entire sphere.
- **L2 spherical cap discrepancy**:
\[
D_{L^2,\text{cap}}(\omega_N)=\left(\int_{C\in\text{cap}}\left|\frac{\#(\omega_N\cap C)}{N}-\frac{\mu(C)}{\mu(S^2)}\right|^2dC\right)^{1/2}
\]
### Main Conclusions
- **Theorem 1.2**: For any parameter selection, there exist two constants $c_1, c_2\in\mathbb{R}^+$ such that
\[
c_1\sqrt{N}\leq D_{\text{sup},\text{cap}}(\diamond(N))\leq c_2\sqrt{N}
\]
- **Corollary 1.3**: For any parameter selection,
\[
D_{L^2,\text{cap}}(\diamond(N))\leq c_2\sqrt{N}
\]
- **Theorem 1.4**: For specific parameter selections (such as $n = 1$ and $r_j=4j$), we have
\[
\frac{1}{\sqrt{N}}+o\left(\frac{1}{\sqrt{N}}\right)\leq D_{\text{sup},\text{cap}}(\diamond(N))<4 + 2\sqrt{2}\cdot\frac{1}{\sqrt{N}}
\]
Through these studies, the author shows the superior performance of Diamond ensemble in the uniformity of the distribution of point sets on the sphere and provides new perspectives and tools for future research.