Colouring of Graphs

Unit III
Vertex Colouring
• Vertex colorings,
• greedy algorithm and its consequences.
• Edge-colorings,
• Vizing theorem on edge-colorings.
we investigate the colouring of graphs and maps, with special reference to the four-colour theorem and
related topics. 
 
A k-vertex colouring of G is an assignment of k colours, 1, 2, ... , k, to the vertices of G; the colouring is
proper if no two distinct adjacent vertices have the same colour.  
Thus a proper k-vertex colouring of a loopless graph G is a partition (VI, V2 , ••• , Vn.) of V into k
(possibly empty) independent sets.  
G is k-vertex-colourable if G has a proper k-vertex colouring. It will be convenient to refer to a 'proper
vertex colouring' as, simply, a colouring and to a 'proper k-vertex colouring' as a k-colouring; 
we shall similarly abbreviate 'k-vertex-colourable' to k-colourable.’  
(G) = 2 if and only if G is a non-null bipartite graph. Note that every tree is 2-colourable, as is any
cycle graph with an even number of vertices. The graphs which are 3-chromatic include the cycle
graphs or wheels with an odd number of vertices and the Petersen graph. The wheels with an even
number of vertices are 4-chromatic. If the graph has n vertices, then its chromatic number cannot
exceed n, and if the graph contains as a subgraph, then its chromatic number cannot be less than r,

THEOREM 17.1. If G is a simple graph with largest vertex-degree    then G is (Δ+l) -coIourable.  
 
The proof is by induction on the number of vertices of G. Let G be a simple graph with n vertices. If
we delete any vertex v and its incident edges, then the graph that remains is a simple graph with
n - 1vertices and largest vertex-degree at most Δ . By our induction hypothesis, this graph is (Δ+l)-
colourable. A (Δ+l)- coiouring for G is then obtained by colouring v with a different colour from the
(at most Δ) vertices adjacent to v. 
 (Brooks, 1941).: If G is a simple connected graph which is not a complete graph, and if the
largest vertex-degree of G is (Δ > 3), then G is Δ colourable.

 The five-colour theorem:.Every simple planar graph is 5-colourable..

 One of the most famous unsolved problems in mathematics - The 'four-colour problem
Every simple planar graph is 4- colourable.

 
Greedy algorithm 
 
Unfortunately, there is no efficient algorithm available for coloring a graph with minimum number of colors as the
problem is a known NP Complete problem. There are approximate algorithms to solve the problem though.
Following is the basic Greedy Algorithm to assign colors. It doesn’t guarantee to use minimum colors, but it
guarantees an upper bound on the number of colors. The basic algorithm never uses more than d+1 colors where d
is the maximum degree of a vertex in the given graph. 

Basic Greedy Coloring Algorithm: 

1. Color first vertex with first color. 
2. Do following for remaining V-1 vertices. 
….. a) Consider the currently picked vertex and color it with the 
lowest numbered color that has not been used on any previously 
colored vertices adjacent to it. If all previously used colors 
appear on vertices adjacent to v, assign a new color to it. 
Greedy algorithm another approach 
 
A very abstract algorithm.  The abstract algorithm we now describe is how you should remember the algorithm.
Knowing this version, you will be able to use it to color a given small graph by hand. You don’t need to see code to
do it. We want to use as few colors as possible. Therefore: it makes sense always to use the smallest possible color
when coloring a vertex. We call this a greedy choice.

Our algorithm, then, is: Color the vertices one by one, as follows:  
To color a vertex, choose the smallest color that is not already the color of a neighbor. We use this algorithm to color
the graph that appears to the right. For clarity, we use vertex names rather than integers.  
1. Give A the color 0.  
2. Give B the color 1, since its colored neighbors have the color 0.
3. Give C the color 2, since its colored neighbors have the colors 0 and 1.
4. Give D the color 0, since its colored neighbors have colors 1 and 2.
5. Give E the color 2, since its colored neighbors have colors 0, 1.
6. Give F the color 1, since its colored neighbors have colors 0, 2.
We want to be estimate the execution time and space requirements of this algorithm. That requires us to go into more
detail on how we can determine the colors of the neighbors of a vertex. We proceed to do this. 
Colouring

In this section we colour the edges of a graph. As we shall see, the four-colour theorem for planar graphs is
equivalent to a theorem concerning edge colourings of cubic maps.
A graph G is k-colourable(e) (or k-edge colourable) if its edges can be coloured with k colours so that no two
adjacent edges have the same colour.
If G is k-colourable but not (k - l)-colourable, we say that the chromatic index of G is k, and write '(G) = k. For
example, Fig. 20.1 shows a graph G for which '(G) = 4.

1. If G is bipartite, then '(G)

THEOREM 20.1 (Vizing, 1964). If G is a simple graph with largest vertex-degree A,then '(G)
If n is even, then we first obtain Kn by joining the complete graph Kn-1 to a single vertex. If we now colour the
edges of Kn_i(Konig 1916). IfG is a bipartite graph with largest vertex-degree A, then x'(G) = A. as above, then
there is one colour missing at each vertex, and these missing colours are all different. We complete the edge
colouring of Kn by colouring the remaining edges with these missing colours .

(Konig 1916). If G is a bipartite graph with largest vertex-degree Δ, then x'(G) = Δ