Choosability

In graph theory, a branch of mathematics, list coloring is a type of graph coloring where each vertex can be restricted to a list of allowed colors. It was first studied in the 1970s in independent papers by Vizing
and by Erdős, Rubin, and Taylor.

Definition

Given a graph and given a set of colors for each vertex (called a list), a list coloring is a ''choice function'' that maps every vertex to a color in the list . As with graph coloring, a list coloring is generally assumed to be proper, meaning no two adjacent vertices receive the same color. A graph is -choosable (or -list-colorable) if it has a proper list coloring no matter how one assigns a list of colors to each vertex. The choosability (or list colorability or list chromatic number) of a graph is the least number such that is -choosable.

More generally, for a function assigning a positive integer to each vertex , a graph is -choosable (or -list-colorable) if it has a list coloring no matter how one assigns a list of colors to each vertex . In particular, if for all vertices , -choosability corresponds to -choosability.

Examples

Consider the complete bipartite graph , having six vertices , , , , , such that and are each connected to all of , , , and , and no other vertices are connected. As a bipartite graph, has usual chromatic number 2: one may color and in one color and , , , in another and no two adjacent vertices will have the same color. On the other hand, has list-chromatic number larger than 2, as the following construction shows: assign to and the lists and . Assign to the other four vertices the lists , , , and . No matter which choice one makes of a color from the list of and a color from the list of , there will be some other vertex such that both of its choices are already used to color its neighbors. Thus, is not 2-choosable.

On the other hand, it is easy to see that is 3-choosable: picking arbitrary colors for the vertices and leaves at least one available color for each of the remaining vertices, and these colors may be chosen arbitrarily.

More generally, let be a positive integer, and let be the complete bipartite graph . Let the available colors be represented by the different two-digit numbers in radix .
On one side of the bipartition, let the vertices be given sets of colors in which the first digits are equal to each other, for each of the possible choices of the first digit .
On the other side of the bipartition, let the vertices be given sets of colors in which the first digits are all distinct, for each of the possible choices of the -tuple
The illustration shows a larger example of the same construction, with .

Then, does not have a list coloring for : no matter what set of colors is chosen for the vertices on the small side of the bipartition, this choice will conflict with all of the colors for one of the vertices on the other side of the bipartition. For instance if the vertex with color set is colored 01, and the vertex with color set is colored 10, then the vertex with color set cannot be colored.
Therefore, the list chromatic number of is at least ..

Similarly, if $n=\tbinom,$ then the complete bipartite graph is not -choosable. For, suppose that colors are available in total, and that, on a single side of the bipartition, each vertex has available to it a different -tuple of these colors than each other vertex. Then, each side of the bipartition must use at least colors, because every set of colors will be disjoint from the list of one vertex. Since at least colors are used on one side and at least are used on the other, there must be one color which is used on both sides, but this implies that two adjacent vertices have the same color. In particular, the utility graph has list-chromatic number at least three, and the graph has list-chromatic number at least four.

Properties

For a graph , let denote the chromatic number and the maximum degree of . The list coloring number satisfies the following properties.
# . A -list-colorable graph must in particular have a list coloring when every vertex is assigned the same list of colors, which corresponds to a usual -coloring.
# cannot be bounded in terms of chromatic number in general, that is, there is no function such that holds for every graph . In particular, as the complete bipartite graph examples show, there exist graphs with but with arbitrarily large.
# where is the number of vertices of .
# .
# if is a planar graph.
# if is a bipartite planar graph.

Computing choosability and (''a'', ''b'')-choosability

Two algorithmic problems have been considered in the literature:
# -''choosability'': decide whether a given graph is -choosable for a given , and
# -''choosability'': decide whether a given graph is -choosable for a given function $f : V \to \$.
It is known that -choosability in bipartite graphs is $\Pi^p_2$-complete for any , and the same applies for 4-choosability in planar graphs, 3-choosability in planar triangle-free graphs, and (2, 3)-choosability in bipartite planar graphs.. For -free graphs, that is, graphs excluding a 5-vertex path graph, -choosability is fixed-parameter tractable.

It is possible to test whether a graph is 2-choosable in linear time by repeatedly deleting vertices of degree zero or one until reaching the 2-core of the graph, after which no more such deletions are possible. The initial graph is 2-choosable if and only if its 2-core is either an even cycle or a theta graph formed by three paths with shared endpoints, with two paths of length two and the third path having any even length.

Applications

List coloring arises in practical problems concerning channel/frequency assignment..

See also

* List edge-coloring

References

Further reading
*, Chapter 34 ''Five-coloring plane graphs''.
*Diestel, Reinhard. ''Graph Theory''. 3rd edition, Springer, 2005. Chapter 5.4 ''List Colouring''.