Independent-Set Threshold Games and Geodetic Removal on Odd Cycles
Overview
Geodetic removal ends when the unselected vertices cannot generate the fixed original graph by repeatedly adjoining shortest paths. On every odd cycle of order at least five, a selected set is terminal exactly when it contains a maximum independent set of a cycle obtained by changing the cyclic order. This correspondence gives a dynamic pairing strategy for the second player, with an exact winning time, and extends to independent-set threshold games on paths and cycles.
Original public questions
Conjecture 5.5: geodetic removing games on odd cycles
Bret J. Benesh, Dana C. Ernst, Marie Meyer, Sarah K. Salmon and Nándor Sieben: Impartial geodetic removing games on graphs — arXiv 2606.05483v1, page 7: Conjecture 5.5, the second-player win in the geodetic removal game on every odd cycle of order at least five.
Original abstract (English)
In the geodetic removal game, players select vertices until the convex hull of the unselected vertices, taken along shortest paths in the original graph, ceases to be the whole graph. We prove that the second player wins on every odd cycle of order at least five, resolving a conjecture of Benesh, Ernst, Meyer, Salmon and Sieben. The proof characterizes terminal selected sets as those containing a maximum independent set of an auxiliary cycle, then uses a dynamic pairing strategy. More generally, consider the impartial game in which vertices are selected until their union contains an independent set of a prescribed size r, with r at least 2. On every path or cycle of order at least 2r, the second player can force termination on exactly move 2r - 2. We classify the remaining feasible path thresholds and prove the same exact-turn result for feasible thresholds on bipartite graphs with a given perfect matching. The strategies maintain paired selected vertices until a final move deliberately breaks the pairing. They admit constant-time responses after linear initialization.
Mathematical review
The principal conclusions have a public kernel-checked proof package and reviewed correspondence with the paper. This is distinct from external peer review.
Review standard