Collinearities in Kinetic Point Sets

Let $P$ be a set of $n$ points in the plane, each point moving along a given trajectory. A {\em $k$-collinearity} is a pair $(L,t)$ of a line $L$ and a time $t$ such that $L$ contains at least $k$ points at time $t$, the points along $L$ do not all coincide, and not all of them are collinear at all times. We show that, if the points move with constant velocity, then the number of 3-collinearities is at most $2\binom{n}{3}$, and this bound is tight. There are $n$ points having $\Omega(n^3/k^4 + n^2/k^2)$ distinct $k$-collinearities. Thus, the number of $k$-collinearities among $n$ points, for constant $k$, is $O(n^3)$, and this bound is asymptotically tight. In addition, there are $n$ points, moving in pairwise distinct directions with different speeds, such that no three points are ever collinear.

💡 Research Summary

The paper investigates the combinatorial behavior of collinearities among points that move with constant velocity in the plane. A “k‑collinearity” is defined as a pair (L, t) where a line L contains at least k points at time t, the points are not all coincident, and they are not permanently collinear (i.e., they do not lie on the same line for all times). The authors focus on three main questions: (1) the maximum possible number of 3‑collinearities, (2) the asymptotic bound for k‑collinearities when k is a fixed constant, and (3) whether it is possible to arrange moving points so that no three ever become collinear.

Geometric model.
Each moving point p is represented by a line Lₚ in three‑dimensional space ℝ³, where the third coordinate is time. For two points a and b, the set Sₐ,ᵦ of all point‑time pairs that are collinear with a and b is the union of all horizontal lines intersecting both Lₐ and Lᵦ. Depending on the relative motion of a and b, Sₐ,ᵦ takes one of three forms: (i) a non‑horizontal plane if a and b have identical velocity vectors, (ii) the union of a non‑horizontal plane and a horizontal plane if they collide, or (iii) a hyperbolic paraboloid (a doubly‑ruled surface of degree two) in the generic case where the trajectories are skew.

A third point c is collinear with a and b at time t precisely when its trajectory L_c meets Sₐ,ᵦ at the point (x, y, t). Since a plane (degree 1) can intersect a line at most once and a hyperbolic paraboloid (degree 2) at most twice, any triple of points that is not permanently collinear can be collinear at most two distinct times. Consequently, each of the C(n, 3) triples contributes at most two 3‑collinearities, yielding the upper bound

  #3‑collinearities ≤ 2·C(n, 3).

Tightness of the bound.
The authors construct a point set that attains this bound. All points start on a unit circle centered at (−1, 1) and move with speed 1 toward the origin, each with a distinct direction θ_i = 3π/2 + π/(4i). As time tends to ±∞ the points lie approximately on concentric circles about the origin, and the cyclic order of any three points reverses, guaranteeing two distinct collinearities for every triple. Small perturbations of positions or directions preserve the property, confirming that the bound 2·C(n, 3) is exact (Theorem 6).

Extension to k‑collinearities.
A k‑collinearity necessarily contains C(k, 3) distinct 3‑collinearities, so the same argument gives

  #k‑collinearities ≤ 2·C(n, 3) / C(k, 3).

For fixed k this is O(n³). To show that this bound is asymptotically tight, the paper presents two constructions.

When n ≥ k²: Points are placed on two parallel vertical lines x = 0 and x = 1. Points on each line are grouped into “layers” with increasing speeds. At integer times a certain number of points on each line collide (share the same y‑coordinate). Each such pair of collisions defines a line that contains k points (the colliding points from both lines), producing a k‑collinearity. Counting over all times yields Ω(n³/k⁴) k‑collinearities.

When k ≤ n < k²: Points are partitioned into groups of size at least ⌈k/2⌉, each group having a simultaneous ⌈k/2⌉‑collision at time 0. Any line joining two such collisions is a k‑collinearity, giving Ω(n²/k²) instances. Together these give a lower bound Ω(n³/k⁴ + n²/k²), matching the upper bound up to constant factors (Theorem 2).

Zero‑collinearity construction.
The authors also demonstrate that it is possible to have n points with pairwise distinct velocities and directions yet never produce a 3‑collinearity. They place each point at distance 1 from the origin with direction θ_i = π/2 + π/(2i) and speed 1. The trajectories form one ruling of a hyperbolic paraboloid; any horizontal slice of this surface is a circle, and a line can intersect a circle in at most two points, so three points can never be collinear. By stretching the configuration in the x‑direction (doubling x‑coordinates and x‑components of velocities) they ensure all speeds become distinct while preserving the non‑collinearity property (Theorem 7).

Implications and future work.
The results give exact combinatorial limits for collinearity events in kinetic point sets with linear motion, which directly translate into worst‑case certificate failure counts for kinetic data structures based on orientation certificates. The tight constructions also illustrate that the worst case can be realized without any permanently collinear triples. Open directions include extending the analysis to higher dimensions, to non‑linear trajectories (e.g., circular or polynomial motion), and to other types of certificates beyond orientation. The techniques linking space‑time geometry (planes, hyperbolic paraboloids) to kinetic events may prove useful for a broader class of kinetic combinatorial problems.