On k-hypertournament losing scores

We give a new and short proof of a theorem on k-hypertournament losing scores due to Zhou et al. [G. Zhou, T. Yao, K. Zhang, On score sequences of k-tournaments, European J. Comb., 21, 8 (2000) 993-1000.]

💡 Research Summary

The paper revisits a classic result on the losing‑score sequences of k‑hypertournaments, originally proved by Zhou, Yao, and Zhang in 2000, and supplies a markedly shorter and more transparent proof. A k‑hypertournament is defined on a vertex set V of size n by assigning a single orientation to every k‑element subset of V; thus each k‑tuple becomes a directed hyperedge. For a vertex v, its losing score L(v) counts the number of hyperedges that contain v in which v is not the winner. Collecting these values and sorting them non‑decreasingly yields the sequence L = (l₁, l₂, …, lₙ).

Zhou et al. showed that L can be realized by some k‑hypertournament if and only if two conditions hold: (1) each entry lies between 0 and (\binom{n-1}{k-1}); (2) for every t (1 ≤ t ≤ n) the partial sum satisfies (\sum_{i=1}^{t} l_i \ge \binom{t}{k}), with equality when t = n. Their original proof relied on a delicate “switch” operation that repeatedly altered orientations to adjust the score sequence, making the argument technically heavy.

The new proof eliminates the switch machinery and rests on two elementary ideas: a minimal‑counterexample contradiction and a constructive insertion process. First, assume a sequence violates condition (2) for some smallest t. Because all smaller prefixes satisfy the inequality, the authors show that the t‑th vertex’s losing score can be increased without breaking the total sum constraint (\sum_{i=1}^{n} l_i = \binom{n}{k}). This uses the identity (\binom{t}{k} = \binom{t-1}{k} + \binom{t-1}{k-1}) to balance the deficit. The contradiction proves that any sequence meeting the two conditions cannot have a violating prefix.

Second, to prove sufficiency, the authors construct a hypertournament inductively. Starting with the smallest entry l₁, they place a vertex v₁ whose losing score is forced to be l₁. Assuming a hypertournament on the remaining n‑1 vertices that realizes the truncated sequence, they add v₁ back by orienting exactly l₁ of the (\binom{n-1}{k-1}) hyperedges that involve v₁ so that v₁ loses in those edges and wins in the rest. This insertion leaves the losing scores of the other vertices unchanged, thereby extending the realization to the full sequence.

By combining the contradiction argument (necessity) with the insertion construction (sufficiency), the paper delivers a concise bidirectional proof: a sequence satisfies the Zhou‑Yao‑Zhang inequalities if and only if it is the losing‑score sequence of some k‑hypertournament. The approach not only shortens the original argument but also clarifies the combinatorial structure underlying k‑hypertournaments.

Beyond the immediate result, the authors discuss implications. The insertion technique is readily adaptable to weighted hyperedges, non‑uniform hyperedge sizes, or dynamic settings where orientations evolve over time. Moreover, analogous arguments could be developed for winning‑score sequences or for other statistics derived from hypergraph orientations. The paper suggests future work on algorithmic verification of score sequences, efficient generation of hypertournaments with prescribed scores, and extensions to multi‑player game models that naturally map onto hypergraph orientations.

In summary, the article provides a clean, constructive proof of the classic characterization of losing‑score sequences in k‑hypertournaments, streamlining the theory and opening avenues for broader generalizations and computational applications.