A Path Variant of the Explorer Director Game on Graphs

The Explorer-Director game, first introduced by Nedev and Muthukrishnan (2008), simulates a Mobile Agent exploring a ring network with an inconsistent global sense of direction. Two players, the Explorer and the Director, jointly control a token’s movement on the vertices of a graph $G$ with initial location $v$. Each turn, the Explorer calls any valid distance, $d$, aiming to maximize the number of vertices the token visits, and the Director moves the token to any vertex distance $d$ away aiming to minimize the number of visited vertices. The game ends when no new vertices can be visited, assuming optimal play, and we denote the total number of visited vertices by $f_d(G,v)$. Here we study a variant where, if the token is on vertex $u$, the Explorer is allowed to select any valid \emph{path length}, $\ell$, and the Director now moves the token to any vertex $v$ such that $G$ contains a $uv$ path of length $\ell$. The corresponding parameter is $f_p(G,v)$. In this paper, we explore how far apart $f_d(G,v)$ and $f_p(G,v)$ can be, proving that for any $n$ there are graphs $G$ and $H$ with $f_p(G,v)-f_d(G,v)>n$ and $f_d(H,v)-f_p(H,v)>n$.

💡 Research Summary

**
The paper studies a natural extension of the Explorer‑Director game originally introduced by Nedev and Muthukrishnan (2008) to model a mobile agent exploring a ring network with an inconsistent global sense of direction. In the original distance‑based version, the Explorer announces a distance d and the Director moves the token from its current vertex u to any vertex v at distance d; the Explorer tries to maximize the total number of distinct vertices visited, while the Director tries to minimize it. The optimal number of visited vertices, given a graph G and a starting vertex v, is denoted f₍d₎(G,v).

The authors propose a “path variant” in which the Explorer announces a path length ℓ instead of a distance. The Director may then move the token to any vertex w such that there exists a u‑w walk of exactly ℓ edges in G. The optimal visited‑vertex count for this variant is denoted f₍p₎(G,v). The central question is how far apart these two parameters can be for the same graph.

First, the paper adapts the notion of a “closed set” (used in the distance version) to a “path‑closed set”: a vertex set U is path‑closed if for every u∈U and every integer ℓ for which a u‑x path of length ℓ exists in G, there is also a vertex y∈U with a u‑y path of the same length. Such a set guarantees that the Director can keep the token forever inside U, regardless of the Explorer’s choices.

The authors then focus on bipanpositionable graphs – a class of Hamiltonian bipartite graphs introduced by Kao (2008) that, for any two vertices x,y and any admissible integer k, contain a Hamiltonian cycle in which the distance between x and y along the cycle equals k. They prove a striking uniform bound: for every bipanpositionable graph G with at least four vertices, f₍p₎(G,v)=4 for any starting vertex v. The proof constructs a C₄ (a 4‑cycle) containing the start vertex and shows that, regardless of the announced path length ℓ (including ℓ=1,2), the Director can always move the token to another vertex of this C₄, thereby confining the game to exactly four vertices.

Since all hypercubes Qₙ (n‑dimensional cubes) are bipanpositionable for n≥2, the corollary follows: f₍p₎(Qₙ,v)=4 for n≥2 (and trivially 1 for n=0, 2 for n=1).

The second part of the paper returns to the original distance variant on hypercubes. Using the known characterization that f₍d₎(G,v) equals the size of a minimum closed set, the authors obtain a lower bound f₍d₎(Qₙ) ≥ n+1, because every vertex has eccentricity n and a unique antipodal vertex at distance n. They also observe that f₍d₎(Qₙ) must be even, as any closed set in a hypercube must contain antipodal pairs.

Upper bounds are derived in two ways.

1. By embedding Qₙ into a 2ⁿ‑vertex cycle C₂ⁿ that preserves distances, the authors show f₍d₎(Qₙ) ≤ 2ⁿ.
2. When n is not a power of two, let p be the smallest odd prime divisor of n; then f₍d₎(Qₙ) ≤ ⌈2ⁿ·(p−1)/p⌉. This improves the trivial 2ⁿ bound for many n.

A recursive bound is also proved: for any integer x with ⌊n/2⌋ ≤ x < n, f₍d₎(Qₙ) ≤ 2·f₍d₎(Qₓ). This allows one to bootstrap known values for smaller dimensions to obtain bounds for larger ones.

Combining these results, the authors demonstrate that the two parameters can diverge arbitrarily. Because f₍p₎ is constant (4) on all bipanpositionable graphs, while f₍d₎ can grow linearly (≥ n+1) or exponentially (≤ 2ⁿ) on hypercubes, for any integer N one can construct a graph G with f₍p₎(G,v) − f₍d₎(G,v) > N and another graph H with f₍d₎(H,v) − f₍p₎(H,v) > N. This establishes that the path variant and the distance variant are fundamentally different in terms of the exploration power they afford the Explorer.

The paper concludes by suggesting several directions for future work: extending the analysis to non‑bipanpositionable graphs, investigating multi‑token or multi‑Explorer settings, and exploring algorithmic aspects of computing f₍d₎ and f₍p₎ for arbitrary graphs. The results deepen our understanding of pursuit–evasion style games on graphs and highlight how subtle changes in the movement rules can dramatically affect the achievable exploration.