A (proper) -edge colouring of a graph is a function such that for . Define the edge-chromatic number or chromatic index as
Recall the definition of line graph of . It follows from definition that . Since has a clique of order if there exists a vertex of degree , from the trivial bound for chromatic number we have that . One can check that and so
The above bound was considerably tightened leading to very rigid behaviour for edge-chromatic numbers.
.
It is non-trivial to determine whether it is or . See [Vinzing Theorem] for more on this theorem, variants and applications. We first state a lemma and prove the theorem. We defer the proof of the Lemma to later.
Let be a vertex such that and its neighbours have at most degree with exactly one neighbour having degree . Then if is -edge colourable so is .
We prove the upper bound by induction. The theorem holds trivially for . Without loss of generality assume that is connected and has strictly more than vertices. Assume that the theorem holds for any graph with fewer vertices or edges.
Let be the vertex with maximum degree i.e., . Let such that . If no such exists then, is a star graph and proving the theorem. Add an additional vertex and connect it to . Call this new graph . Observe that in , satisfies the conditions of Lemma 11.14 with . Furthermore has fewer edges than and so . Thus by Lemma 11.14, we have that is -edge colourable and so is . ∎
An alternate proof of the theorem using edge-deletions but nevertheless using some of the ideas in proof of the Lemma 11.14 can be found in [Vinzing Theorem].
We shall give a sketch of proof. For full details, see [Sudakov 2019, Chapter 8].
The proof is by induction on . Let . Then has exactly one neighbour and is an isolated edge. If is -edge colourable, then easily we can colour the isolated edge by the same colour and get a -edge colouring of .
Let . By adding additional vertices if necessary, we can assume without loss of generality that and there exists such that and for all .
Let be a -edge colouring of . Let be the number of neighbours of missed by colour . One neighbour misses only one colour and rest of the neighbours miss colours. So by double counting, we have that
Among all -edge colourings choose one that minimizes . Suppose it holds for such a colouring that for all
(11.1) |
Then since is odd, there exists . Without loss of generality, we can assume that . Set . Let be the graph obtained from by deleting and all edges of colour . So has a -edge colouring. By induction hypothesis, we can verify that has a -edge colouring. Now putting back the edges of colour as well as colouring with colour , we get a -colouring of . Note that other neighbours of have other missing colours that can be used.
We are left to prove (11.1). WLOG, let and assume . Define to be the sub-graph formed by edges of colour or . Observe that consists of path or even cycle components. Since internal vertices in a path component and all vertices in a cycle component are missed by both the colours, the difference between and arises from a path component, say . starts and ends with a vertex in . Now interchanging the edge colours along , we get a new colouring with , and the rest of ’s remain unchanged. So trivially,
which contradicts the minimality of . Thus and thus (11.1) holds. ∎