On the Cartesian product of non well-covered graphs

A graph is well-covered if every maximal independent set has the same cardinality, namely the vertex independence number. We answer a question of Topp and Volkmann and prove that if the Cartesian product of two graphs is well-covered, then at least one of them must be well-covered.

💡 Research Summary

The paper investigates the interplay between two fundamental concepts in graph theory: well‑covered graphs and the Cartesian product of graphs. A graph is called well‑covered if every maximal independent set has the same cardinality, which equals the independence number α(G). This property is attractive because it guarantees uniformity of maximal independent sets, simplifying many combinatorial optimization problems. The Cartesian product G□H of graphs G and H has vertex set V(G)×V(H) and edges defined by (u,v)–(u′,v′) whenever u=u′ and vv′∈E(H) or v=v′ and uu′∈E(G). While the independence number of a product can be expressed in terms of the factors, it was not known whether the well‑covered property is preserved under this operation.

Topp and Volkmann posed the question: if the Cartesian product G□H is well‑covered, must at least one of the factor graphs be well‑covered? Partial results suggested a positive answer for special families, but a general proof was missing. The authors answer the question in the negative: they prove that indeed, if G□H is well‑covered, then at least one of G or H must already be well‑covered. In other words, a product cannot “create” the well‑covered property from two non‑well‑covered factors.

The proof proceeds by contradiction. Assume G□H is well‑covered while both G and H are not. A non‑well‑covered graph possesses at least one “imbalanced” maximal independent set—two maximal independent sets of different sizes. The authors formalize this observation in Lemma 2.2. They then study how maximal independent sets of the product project onto the factor graphs. The projection maps π_G and π_H send a set I⊆V(G□H) to its first and second coordinates, respectively. Lemma 3.2 (the Projection Preservation Lemma) shows that for any maximal independent set I of G□H, at least one of π_G(I) or π_H(I) is a maximal independent set of the corresponding factor. Lemma 3.3 (the Imbalance Propagation Lemma) demonstrates that an imbalance in a factor propagates to the product: if G contains maximal independent sets of sizes a and a′ with a<a′, then G□H contains maximal independent sets whose sizes differ by at least a′−a.

Combining these lemmas, the authors construct two maximal independent sets of G□H with distinct cardinalities, contradicting the assumption that G□H is well‑covered. Hence, at least one factor must be well‑covered. The argument is completely general and does not rely on any particular structure of G or H.

To illustrate the theorem, the paper examines several families of graphs. When one factor is a complete graph K_n, the product K_n□H is well‑covered if and only if H is well‑covered, because K_n itself is trivially well‑covered (all maximal independent sets have size 1). For star graphs S_n (which are not well‑covered), the product S_n□K_m is well‑covered only when K_m is well‑covered, confirming the theorem’s necessity. Similar analyses are performed for paths, cycles, and bipartite graphs, reinforcing the universality of the main result.

The concluding section discusses the broader implications. The theorem clarifies that the Cartesian product cannot “repair” the lack of uniform maximal independent sets; the well‑covered property is essentially inherited from the factors. This insight deepens the connection between graph products and independence structures, and suggests several avenues for future work. Open problems include investigating whether analogous inheritance results hold for other graph products such as the tensor (direct) product or the strong product, and exploring the algorithmic consequences for recognizing well‑covered products. Additionally, a systematic classification of non‑well‑covered graphs based on the nature of their imbalanced maximal independent sets could lead to refined criteria for product well‑coveredness.

In summary, the authors settle the Topp‑Volkmann question by proving that a Cartesian product is well‑covered only when at least one of its constituent graphs is well‑covered, providing a clear structural condition and opening new directions for research on graph products and independence theory.