Strong ε-Net: Bounds & Constructions

• Strong ε-nets are subsets of a finite point set in ℝ^d ensuring any geometric range containing more than an ε fraction of points includes at least one point from the subset.
• They differ from weak ε-nets by requiring the net’s points to belong to the original set, making them crucial for range families like axis-parallel rectangles, halfspaces, and disks.
• Research in this field provides nearly tight upper and lower bounds using techniques such as slicing, grid partitioning, and Delaunay triangulation for effective net constructions.

A strong $\epsilon$-net is a combinatorial geometric object that provides a highly constrained representative subset of a finite point set in $\mathbb{R}^d$ with respect to a specified family of geometric ranges. For a given finite set $P$ of $n$ points and a family $\mathcal{C}$ of (typically convex) subsets of $\mathbb{R}^d$, a strong $\epsilon$-net is a subset $N \subseteq P$ such that every range $C \in \mathcal{C}$ containing more than $\epsilon n$ points of $\mathbb{R}^d$0 also contains at least one point of $\mathbb{R}^d$1. The defining property of a strong $\mathbb{R}^d$2-net, in contrast with weak $\mathbb{R}^d$3-nets, is that its constituent elements are required to be points of $\mathbb{R}^d$4 itself rather than arbitrary points in the ambient space. The article "Small Strong Epsilon Nets" (Ashok et al., 2012) introduces and analyzes small strong $\mathbb{R}^d$5-nets, providing exact and nearly tight upper and lower bounds on their minimal size and covering fraction for key geometric range families in the plane and higher dimensions.

1. Formal Definitions

Let $\mathbb{R}^d$6 be a finite set of $\mathbb{R}^d$7 points and $\mathbb{R}^d$8 a family of geometric ranges. The central notions are:

• Strong $\mathbb{R}^d$9-net: A subset $P$0 is a strong $P$1-net for $P$2 if for every $P$3 with $P$4, the intersection $P$5 is nonempty.
• Weak $P$6-net: As above, but $P$7 need only be a subset of $P$8, not necessarily of $P$9.
• Centerpoint and Strong Centerpoint: A centerpoint $n$0 of $n$1 is a point lying in every convex set containing more than $n$2 points of $n$3; a strong centerpoint for $n$4 is $n$5 contained in every $n$6 with $n$7 for some constant $n$8. In this language, a strong centerpoint is a one-point strong $n$9-net.
• Small strong $\mathcal{C}$0-nets: For a family $\mathcal{C}$1 of ranges, define $\mathcal{C}$2 as the smallest $\mathcal{C}$3 such that every $\mathcal{C}$4-point set $\mathcal{C}$5 admits a strong $\mathcal{C}$6-net $\mathcal{C}$7 of size $\mathcal{C}$8, i.e., $\mathcal{C}$9.

These definitions set the stage for the quantitative analysis of how well a finite subset of $\mathbb{R}^d$0 can represent all large-range subsets defined by $\mathbb{R}^d$1.

2. Main Results for Core Geometric Range Families

Three principal families are examined in the plane: axis-parallel rectangles ($\mathbb{R}^d$2), halfspaces ($\mathbb{R}^d$3), and disks ($\mathbb{R}^d$4). The paper provides explicit finite-size bounds and asymptotic rates for $\mathbb{R}^d$5 for each:

(I) Axis-parallel Rectangles $\mathbb{R}^d$6

• Strong Centerpoint: For axis-parallel boxes in $\mathbb{R}^d$7, $\mathbb{R}^d$8. In $\mathbb{R}^d$9, this specializes to $\epsilon$0.
• Small Nets in $\epsilon$1: For $\epsilon$2, explicit lower (LB) and upper (UB) bounds for $\epsilon$3 are as follows:

$\epsilon$4	1	2	3	4	5	6	7	8	9	10
LB	$\epsilon$5	$\epsilon$6	$\epsilon$7	$\epsilon$8	$\epsilon$9	$N \subseteq P$0	$N \subseteq P$1	$N \subseteq P$2	$N \subseteq P$3	$N \subseteq P$4
UB	$N \subseteq P$5	$N \subseteq P$6	$N \subseteq P$7	$N \subseteq P$8	$N \subseteq P$9	$C \in \mathcal{C}$0	$C \in \mathcal{C}$1	$C \in \mathcal{C}$2	$C \in \mathcal{C}$3	$C \in \mathcal{C}$4

• Asymptotic Bounds: $C \in \mathcal{C}$5 for all $C \in \mathcal{C}$6; $C \in \mathcal{C}$7 by recursive grid and slicing constructions.

(II) Halfspaces $C \in \mathcal{C}$8 in $C \in \mathcal{C}$9

• No Strong Centerpoint: $\epsilon n$0.
• Upper Bounds: $\epsilon n$1 for all $\epsilon n$2.
• Lower Bounds: For odd $\epsilon n$3, $\epsilon n$4 (tight); for even $\epsilon n$5, $\epsilon n$6. Notably, $\epsilon n$7, $\epsilon n$8.

(III) Disks $\epsilon n$9 in $\mathbb{R}^d$00

• No Strong Centerpoint: $\mathbb{R}^d$01.
• Upper Bound (for $\mathbb{R}^d$02): $\mathbb{R}^d$03.
• Lower Bounds: For odd $\mathbb{R}^d$04, $\mathbb{R}^d$05; for even $\mathbb{R}^d$06, $\mathbb{R}^d$07. For $\mathbb{R}^d$08, there is an explicit construction yielding $\mathbb{R}^d$09.

3. Fundamental Proof Techniques

Several foundational methodologies underpin the analysis and construction of small strong $\mathbb{R}^d$10-nets:

• Slicing Arguments (Rectangles): Partition $\mathbb{R}^d$11 using one or two coordinate-aligned slabs to localize high-density regions and recurse, yielding moderate-sized nets.
• Grid Arguments (Theorem 3.1): Partition $\mathbb{R}^d$12 into $\mathbb{R}^d$13 horizontal and $\mathbb{R}^d$14 vertical strips; select near-boundary neighbors to block empty rectangles, then apply recursion.
• Convex Hull "Necklace" (Halfspaces): Traverse the convex hull of $\mathbb{R}^d$15, selecting hull vertices so that each large exterior halfspace must contain a selected point.
• Delaunay/Empty-Circle Argument (Disks): Use the Delaunay triangulation around a centerpoint and the crossing count of segments through triangle edges to force that large disks intersect certain key points.
• Recursive Lower Bound: Construct two well-separated hard instances and combine, showing $\mathbb{R}^d$16 for any convex family $\mathbb{R}^d$17 and $\mathbb{R}^d$18.

4. Explicit Constructions

Theoretical bounds are realized via concrete algorithmic constructions for small strong $\mathbb{R}^d$19-nets:

• Axis-parallel Rectangles: Starting from a strong centerpoint $\mathbb{R}^d$20 (with covering ratio $\mathbb{R}^d$21), draw vertical and horizontal lines through $\mathbb{R}^d$22, partitioning $\mathbb{R}^d$23 into four slabs. If any slab contains a large fraction, place a net there and its complement; if not, recursions on two slabs of half-size produce higher-order nets while optimizing the balance parameter $\mathbb{R}^d$24.
• Halfspaces: Maintain a set starting at a hull vertex, walk along the hull, and, upon encountering a halfspace that would avoid all previously selected points and contain too many points, add the current vertex. Ensure at most $\mathbb{R}^d$25 points are selected.
• Disks: Compute a centerpoint $\mathbb{R}^d$26 and the Delaunay triangulation. The triangle containing $\mathbb{R}^d$27 (say, $\mathbb{R}^d$28) provides an edge (e.g., $\mathbb{R}^d$29) heavily crossed by centerpoint-complement segments. The points $\mathbb{R}^d$30 and $\mathbb{R}^d$31 form the net, as any disk avoiding them can only capture a limited fraction of $\mathbb{R}^d$32.

5. Tabulation of Bounds and Asymptotics

Key ranges and their small-$\mathbb{R}^d$33 behaviors in the plane are summarized:

$\mathbb{R}^d$34	$\mathbb{R}^d$35	$\mathbb{R}^d$36	$\mathbb{R}^d$37
$\mathbb{R}^d$38	$\mathbb{R}^d$39	$\mathbb{R}^d$40	$\mathbb{R}^d$41
$\mathbb{R}^d$42	$\mathbb{R}^d$43	$\mathbb{R}^d$44	$\mathbb{R}^d$45
$\mathbb{R}^d$46	$\mathbb{R}^d$47	$\mathbb{R}^d$48	$\mathbb{R}^d$49
Asymptotics	$\mathbb{R}^d$50	$\mathbb{R}^d$51 (odd $\mathbb{R}^d$52), $\mathbb{R}^d$53 even $\mathbb{R}^d$54	$\mathbb{R}^d$55, upper bounds similar to halfspaces (except $\mathbb{R}^d$56)

These results yield the first nontrivial upper and lower bounds for constant-size strong $\mathbb{R}^d$57-nets for the considered range families.

6. Open Problems and Research Directions

Several directions remain open for exploration:

• Determining the exact values of $\mathbb{R}^d$58 for $\mathbb{R}^d$59, $\mathbb{R}^d$60 for even $\mathbb{R}^d$61, and $\mathbb{R}^d$62 for $\mathbb{R}^d$63.
• Improving upper bounds for disks beyond $\mathbb{R}^d$64 and closing the gap between bounds of type $\mathbb{R}^d$65 (from halfspace-style constructions) and the currently known lower bounds.
• Extending strong $\mathbb{R}^d$66-net bounds to other low-VC-dimension families such as pseudo-disks or convex $\mathbb{R}^d$67-gons (Ashok et al., 2012).

The study of small strong $\mathbb{R}^d$68-nets delineates the precise interaction between geometric range families and discrete point set coverage, distinguishing itself from the theory of weak $\mathbb{R}^d$69-nets by the requirement that covering points lie in $\mathbb{R}^d$70. This work establishes existence results for strong centerpoints of axis-parallel boxes in arbitrary dimension and opens avenues for understanding combinatorial geometry through the lens of point set representativity and geometric range complexity.