EN FR
EN FR


Section: New Results

Advances in Graph Theory

Participants : Miguel Couceiro, Amedeo Napoli, Chedy Raïssi, Jean-Sébastien Sereni, Mario Valencia.

Keywords:

graph theory, extremal graph theory, chromatic number, triangle-free graph, planar graph, graph coloring

We announced in the last report that we started to work on a conjecture by Heckman and Thomas from 1999. We managed to confirm the conjecture and the demonstration was published in January 2014. A classical result by Staton, from 1979, states that every triangle-free graph G with maximum degree at most 3 contains an independent set of order at least 5n/14, where n is the number of vertices of G. Heckman and Thomas conjectured a stronger fact: the fractional chromatic number of such a graph is at most 14/5. We confirmed their conjecture by establishing the following stronger assertion: for any assignment of weights (i.e., real numbers) to the vertices of such a graph G, there exists an independent set I such that the weights of the vertices in I is at least 5/14 times the total weight of the G.

Exploring further the methods we introduced to solve this conjecture, we obtained new results concerning the fractional chromatic number of planar triangle-free graphs. While the fractional chromatic number of such graphs is at most 3 (because their chromatic number is), a construction of Jones proved the existence of triangle-free planar graphs with fractional chromatic number arbitrarily close to 3. Thus one wonders whether there could be such graphs with fractional chromatic number exactly 3. We demonstrated this not to be the case, by proving a general upper bound of 9n3n+1=3(1-13n+1) for every triangle-free planar graph G with n vertices. This bound is qualitatively the best possible: Jones's construction yields graphs with fractional chromatic number 3-cn for some constant c. In addition, a tight bound was obtained if the graphs considered are furthermore required to have maximum degree at most 4. In this case, the bound becomes 3n3n+1.

Motivated by frequency assignment in office blocks, we study the chromatic number of the adjacency graph of a 3-dimensional parallelepiped arrangement. In the case each parallelepiped is within one floor, a direct application of the Four-Colour Theorem yields that the adjacency graph has chromatic number at most 8. We provide an example of such an arrangement needing exactly 8 colors. We also discuss bounds on the chromatic number of the adjacency graph of general arrangements of 3-dimensional parallelepipeds according to geometrical measures of the parallelepipeds (side length, total surface area or volume).