@Article{Czap2015, author = "Czap, JĂșlius", title = "A note on M2-edge colorings of graphs", journal = "Opuscula Math.", year = "2015", volume = "35", number = "3", pages = "287--291", doi = "http://dx.doi.org/10.7494/OpMath.2015.35.3.287", url = "http://dx.doi.org/10.7494/OpMath.2015.35.3.287", abstract = "An edge coloring \(\varphi\) of a graph \(G\) is called an \(M_2\)-edge coloring if \(|\varphi(v)|\le2 \) for every vertex \(v\) of \(G\), where \(\varphi(v)\) is the set of colors of edges incident with \(v\). Let \(K_2(G)\) denote the maximum number of colors used in an \(M_2\)-edge coloring of \(G\). Let \(G_1\), \(G_2\) and \(G_3\) be graphs such that \(G_1\subseteq G_2\subseteq G_3\). In this paper we deal with the following question: Assuming that \(K_2(G_1)=K_2(G_3)\), does it hold \(K_2(G_1)=K_2(G_2)=K_2(G_3)\)?", keywords = "edge coloring, graph", issn = "1232-9274" }