On the directions occurring in lattice-line coverings of the integer plane

Technical Summary: Directions Occurring in Lattice-Line Coverings of the Integer Plane

Problem Statement
The paper investigates the problem of covering the integer lattice $\mathbb{Z}^2$ with a family $\mathcal{F}$ of lines subject to a specific validity constraint: no two lines of different directions in $\mathcal{F}$ may intersect at a lattice point. While lines of different directions are permitted to cross, such crossings must occur strictly off the lattice.

The central question is: Which sets of line directions can occur in such a valid covering family?
The paper restricts its scope to lattice lines (lines containing at least two, and thus infinitely many, lattice points), which are equivalent to lines of rational direction. Without this restriction, the problem is trivial (using irrational slopes ensures no two lines share a lattice point), but the restriction to rational directions introduces significant combinatorial rigidity.

Methodology and Construction
The author constructs a valid covering family $\mathcal{F}$ whose set of realized directions is dense in the space of all directions $\mathbb{RP}^1$. The construction is a recursive process that partitions $\mathbb{Z}^2$ into nested rank-2 sublattice cosets.

1. Rigidity and Incompatibility (Lemma 8):
   The paper establishes that two distinct primitive directions $d_1 = (p_1, q_1)$ and $d_2 = (p_2, q_2)$ are "incompatible" if their determinant $\Delta = p_1q_2 - p_2q_1$ has absolute value 1 (i.e., they are Farey neighbors). If $|\Delta|=1$, any lattice line of direction $d_1$ and any of direction $d_2$ will intersect at a lattice point regardless of their offsets. Thus, a valid family cannot contain both directions.

2. Coset Decomposition (Lemma 9 & 11):
   The core mechanism involves splitting a current region $R$ (a rank-2 coset defined by congruences $x \equiv x_0 \pmod{n_1}, y \equiv y_0 \pmod{n_2}$) into finer sub-cosets.

   • Given coprime integers $s, t$ satisfying specific gcd conditions with $n_1, n_2$, the coset $R$ is partitioned into $|s||t|$ sub-cosets.
   • Crucially, each sub-coset $R_{u_0, w_0}$ is itself a lattice line coset for a new primitive direction $(P, Q) = (n_1s, n_2t)$.
   • This allows the author to assign a specific direction to cover all points in a chosen sub-coset without creating forbidden intersections with lines from other directions, provided the sub-cosets are disjoint.

3. The Steering Lemma (Lemma 14):
   To ensure the set of directions is dense, the construction must be able to realize a direction arbitrarily close to any target angle $\theta$ at every stage of the recursion. The Steering Lemma proves that for any coprime $n_1, n_2$ and any target $\theta$, there exist integers $s, t$ (satisfying the splitting conditions) such that the direction of $(n_1s, n_2t)$ is within $\epsilon$ of $\theta$. This relies on elementary sieve bounds to find suitable $s, t$ in a sufficiently large interval.

4. Recursive Algorithm:

   • Start with $R_0 = \mathbb{Z}^2$.
   • Fix an enumeration of $\mathbb{Z}^2$ ($z_k$) and a dense sequence of target angles ($\theta_k$).
   • At stage $k$, use the Steering Lemma to find a direction close to $\theta_k$ that splits the current region $R_{k-1}$.
   • Partition $R_{k-1}$ into sub-cosets. Reserve one sub-coset (the "lexicographically least" one avoiding $z_k$ if $z_k$ is still uncovered) to be the input for the next stage ($R_k$).
   • Assign the chosen direction to cover all other sub-cosets.
   • Repeat.

Key Results

• Theorem 7: There exists a valid covering family $\mathcal{F}$ of lattice lines such that the set of directions realized by $\mathcal{F}$ is dense in $\mathbb{RP}^1$.
• Coverage: The recursive construction ensures that every point $z \in \mathbb{Z}^2$ is covered by exactly one stage of the recursion. Because the reserved sub-coset shrinks strictly and the enumeration ensures every point is eventually the "target" $z_k$, the intersection of all reserved regions is empty, guaranteeing full coverage.
• Validity: The construction guarantees that lines of different directions never intersect at a lattice point. This is achieved because lines of a specific stage $k$ are confined to the sub-cosets they cover, which are disjoint from the reserved region $R_k$ passed to future stages. Future stages operate on a subset of $R_k$, ensuring geometric separation from previous stages' lines.

Significance and Claims
The paper claims to resolve the question of direction sets for lattice-line coverings by demonstrating that the set of directions need not be finite or even countable in a restrictive sense; it can be made dense.

• Contrast with Trivial Cases: The result highlights the non-triviality of the "rational direction" constraint. While a single direction suffices for a disjoint covering (Example 6), allowing crossings off-lattice enables a dense set of directions, provided the "unimodular" (Farey neighbor) incompatibility is navigated.
• Methodological Contribution: The paper introduces a recursive splitting technique using rank-2 cosets and a "steering" mechanism to control the geometry of the covering.
• Open Problem: The author explicitly poses the generalization of this result to $\mathbb{Z}^n$ for $n \ge 3$ as an open problem (Open Problem 18), noting that the formulation would involve "lattice hyperplanes" or "lattice lines" with similar non-intersection constraints.

Note on Authorship and Verification
The manuscript explicitly discloses that the results, including the reformulation, lemmas, and recursive construction, were discovered through an interactive session between the author and the AI model Claude. The paper emphasizes that the construction was verified through:

1. Brute-force numerical simulations on finite windows.
2. Independent review by a second AI agent.
3. A formal machine-checked proof in Lean (Mathlib).
   The author retains full responsibility for the final exposition and any remaining errors.