A New Construction of Forests with Low Visibility

Kirill Kashkan

2024-06-30

Abstract:A set of points with finite density is constructed in $\mathbb{R}^d$, with $d\geq2$, by adding points to a Poisson process such that any line segment of length $O\left(\varepsilon^{-(d-1)}\ln\varepsilon^{-1}\right)$ in $\mathbb{R}^d$ will contain one of the points of the set within distance $\varepsilon$ of it. The constant implied by the big-$O$ notation depends on dimension only.

Metric Geometry,Combinatorics,Probability

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to construct a dense forest with low visibility, that is, in the high - dimensional space $\mathbb{R}^d$, how to construct a point set $F\subset\mathbb{R}^d$ such that the visibility in any direction is as small as possible under a certain scale while maintaining the finite density of the point set.

Specifically, the paper aims to find a method to construct a point set $F$ in $\mathbb{R}^d$ such that for any $x\in\mathbb{R}^d$, $v\in S^{d - 1}$ (a vector on the unit sphere), and any $\varepsilon>0$, looking from the point $x$ in the direction of $v$, there is always a point $y\in F$ within a distance not exceeding $V(\varepsilon)$ from $x$, and $V(\varepsilon)\in O\left(\varepsilon^{-(d - 1)\ln\varepsilon^{-1}}\right)$.

### Main contributions of the paper

1. **Problem background**:
- This problem can be traced back to Pólya's Orchard Problem, that is, how large the trunk radius should be when trees are evenly placed on the circumference so as to completely block the line of sight from the center.
- Bishop partially solved this problem in his early work and gave a construction in the two - dimensional plane with a visibility of $O(\varepsilon^{-4})$.
- Subsequent researchers such as Alon, Adiceam and Tsokanos have continuously improved this result and proposed constructions in higher dimensions and increased the upper limit of visibility.

2. **Main results**:
- The paper proposes a new probabilistic construction method to reduce visibility by adding additional points in the Poisson point process.
- Specifically, the author proves that there exists a dense forest in the $d$-dimensional space with a visibility $V(\varepsilon)\in O\left(\varepsilon^{-(d - 1)\ln\varepsilon^{-1}}\right)$.
- The key to this construction method lies in discretizing $\varepsilon$ at different scales and adding a sufficiently small number of points at each scale to ensure that the overall density is still finite.

3. **Technical details**:
- Use John's theorem to transform the visibility problem into a probability problem of empty boxes.
- Introduce the concept of test boxes, and determine the positions where points need to be added through these test boxes.
- By calculating the number of test boxes and the probability of empty boxes at different scales, it is finally concluded that the expected number of added points is finite.

### Conclusion

The paper not only provides a new construction method, but also shows how to minimize visibility by optimizing the parameters of the Poisson point process. In addition, the author discusses the connection between this problem and other related problems (such as the Danzer set problem and the minimum dispersion problem) and points out the direction for future research.

In summary, the main goal of this paper is to construct a dense forest with better visibility by improving existing methods, thereby promoting the understanding of such geometric structures.