Trees and co-trees in planar 3-connected graphs An easier proof via Schnyder woods
Christian Ortlieb, Jens M. Schmidt
TL;DR
The paper addresses Grünbaum's conjecture, which asks for a spanning tree in every $3$-connected planar graph whose co-tree also has maximum degree $3$. It offers an easier, Schnyder-wood–driven proof of a related result due to Biedl, showing that a spanning tree can be chosen so that both it and its co-tree have maximum degree at most $5$. The approach builds a graph $H(G)$ from a Schnyder wood on the suspension $G^\sigma$ and its dual $H^\circ(G^*)$, proves degree bounds and connectivity, and then applies a minimum-weight spanning-tree argument in the style of Biedl to obtain the desired tree and co-tree properties. The method uses dual Schnyder woods and ordered path partitions to relate primal and dual structures, offering a streamlined framework that could shine light on Grünbaum's conjecture.
Abstract
Let $G$ be a 3-connected planar graph. Define the co-tree of a spanning tree $T$ of $G$ as the graph induced by the dual edges of $E(G)-E(T)$. The well-known cut-cycle duality implies that the co-tree is itself a tree. Let a $k$-tree be a spanning tree with maximum degree $k$. In 1970, Grünbaum conjectured that every 3-connected planar graph contains a 3-tree whose co-tree is also a 3-tree. In 2014, Biedl showed that every such graph contains a 5-tree whose co-tree is a 5-tree. In this paper, we present an easier proof of Biedl's result
