PlanetMath: Vizing's theorem

(more info)

Math for the people, by the people.

Main Menu

sections Encyclopædia
Papers
Books
Expositions

meta Requests (200)
Orphanage
Unclass'd
Unproven (490)
Corrections (6)
Classification

talkback Polls
Forums
Feedback
Bug Reports

downloads Snapshots
PM Book

information News
Docs
Wiki
ChangeLog
TODO List
Legalese
About

Entry average rating: No information on entry rating

Vizing's theorem

(Theorem)

Definitions and notation

In a graph or multigraph , let $\rho(\hbox{\sc v})$ denote the valency of vertex (node) $\hbox{\sc v}$ , and let $\rho(G)$ denote the largest valency in (often written just $\rho$ on its own, $\Delta(G)$ is another common notation).

Let the multiplicity $\mu(\hbox{\sc v},\hbox{\sc w})$ of vertices (nodes) $\hbox{\sc v}$ and $\hbox{\sc w}$ be the number of parallel edges that link them, and here too let $\mu(G)$ be the largest multiplicity in . A graph is a multigraph for which $\mu(G)=1$ .

An edge coloring of a (multi)graph is a mapping from the set of its edges to a set of items called colors, in such a way that at any vertex $\hbox{\sc v}$ , the $\rho(\hbox{\sc v})$ edges there all have a different color. An edge--coloring is an edge-coloring where $\vert K\vert=k$ .

Note that a loop (an edge joining $\hbox{\sc v}$ to itself) accounts for two of the $\rho(\hbox{\sc v})$ edges at $\hbox{\sc v}$ and cannot have a color different from itself, so edge-colorings as defined here do not exist for pseudographs (structures that are allowed to have loops).

The chromatic index (aka edge-chromatic number) ${\chi\,}'(G)$ is the smallest number for which an edge--coloring of exists. This now standard notation is analogous to the chromatic number $\chi(G)$ which refers to vertex coloring.

Vizing's theorem

For any multigraph (this includes graphs), we have of course

$\begin{displaymath} {\chi\,}'(G) \;\ge\; \rho(G) \end{displaymath}$

immediately from the definition. Surprisingly though, we also have

Theorem (Vizing) $\quad$ For any graph ,

$\begin{displaymath} {\chi\,}'(G) \;\le\; \rho(G)+1 \end{displaymath}$

This celebrated theorem was proved by V. G. Vizing in 1964 while still a graduate student (at Novosibirsk). It is a special case of

Theorem (Vizing) $\quad$ For any multigraph ,

$\begin{displaymath} {\chi\,}'(G) \;\le\; \rho(G)+\mu(G) \end{displaymath}$

(edge-coloring of multigraphs is the subject of Vizing's doctoral thesis, 1965). The theorem was proved independently by R. P. Gupta in 1966; nowadays various versions of the proofs exist. A key rôle is played by connected subgraphs

of colors

and

that cannot be extended further. They are analogous to Kempe chains in face or vertex colorings, but have a simpler structure: they are either paths or closed paths (the latter of even length). See here for a proof (in the case of graphs).

Ore [Ore67] gives a sharper bound for multigraphs. Let the enlarged valency $\rho^+(\hbox{\sc v})$ be given by

$\begin{displaymath} \rho^+(\hbox{\sc v}) \;\mathrel{\mathop{=}\limits^{\smash{\h... ...ho(\hbox{\sc v}) + \max_{\rm w} \mu(\hbox{\sc v},\hbox{\sc w}) \end{displaymath}$

where $\hbox{\sc w}$ can be taken over all vertices of the graph; it effectively only ranges over those adjacent to $\hbox{\sc v}$ (as $\mu(\hbox{\sc v},\hbox{\sc w})$ is zero for the others). Again, let $\rho^+(G)$ denote the maximum enlarged valency occurring in

. Now

Theorem (Ore) $\quad$ For any multigraph ,

$\begin{displaymath} {\chi\,}'(G) \;\le\; \rho^+(G) \end{displaymath}$

Shannon's theorem

The following theorem was proven in 1949 by Claude E. Shannon. He also gives examples, for any value of $\rho$ , of multigraphs that actually attain the bound, the so-called Shannon graphs: they have three vertices, with $\mu(\hbox{\sc a},\hbox{\sc b})=\mu(\hbox{\sc a},\hbox{\sc c})=\lfloor\rho/2\rfloor$ and $\mu(\hbox{\sc b},\hbox{\sc c})=\lceil\rho/2\rceil$ .

Theorem (Shannon) $\quad$ For any multigraph ,

$\begin{displaymath} {\chi\,}'(G) \;\le\; \lfloor\mathord{\mathchoice {\T{3\over 2}} {\T{3\over 2}} {\S{3\over 2}} {\SS{3\over 2}}}\rho(G)\rfloor \end{displaymath}$

While giving a much worse bound for graphs, it gives for some multigraphs a better bound than Vizing's theorem. Nevertheless, it is also possible to prove it from the latter [FW77].

Only in the context of graph colorings is Shannon's theorem understood to refer to the one here; in the wider world the term tends to refer to any of his fundamental theorems in information theory.

Here too Ore [Ore67] gives a sharper bound based on the maximum of a local expression. Let $\sigma(\hbox{\sc v})$ be 0 if $\hbox{\sc v}$ has fewer than two neighbours, and otherwise

$\begin{displaymath} \sigma(\hbox{\sc v}) \;\mathrel{\mathop{=}\limits^{\smash{\h... ...ho(\hbox{\sc v}) + \rho(\hbox{\sc v}') + \rho(\hbox{\sc v}'')) \end{displaymath}$

where $\hbox{\sc v}'$ and $\hbox{\sc v}''$ range over all pairs of distinct neighbours of $\hbox{\sc v}$ . Again, let $\sigma(G)$ be its largest value in the graph.

Theorem (Ore) $\quad$ For any multigraph ,

$\begin{displaymath} {\chi\,}'(G) \;\le\; \max(\rho(G),\lfloor\mathord{\mathchoic... ...\T{1\over 2}} {\S{1\over 2}} {\SS{1\over 2}}}\sigma(G)\rfloor) \end{displaymath}$

Chromatic class

Vizing's theorem has the effect of placing each multigraph in one of the classes 0, 1, 2, ... $\mu(G)$ where the class number is ${\chi\,}'(G)-\rho(G)$ .

For graphs it means they split into just two classes: those that can be edge-colored in $\rho(G)$ colors and those that need $\rho(G)+1$ colors. The logical name for them would be class 0 and class 1; unfortunately the standard terminology is class 1 and 2 (or I and II).

Class I (graphs that can be edge- $\rho$ -colored) contains among others

single -cycles for even $^\dagger$
complete graphs for even $^\ddagger$
bipartite graphs (this is Konig's theorem, 1916)
bridgeless planar trivalent graphs (four-color theorem via Tait coloring)
planar graphs with $\rho(G)\ge8$ (by another theorem of Vizing)
planar graphs with $\rho(G)\ge6$ ?? (conjecture of Vizing)

Class II (graphs that need $\rho+1$ colors) contains among others

single -cycles for odd $^\dagger$
complete graphs for odd $^\ddagger$
for odd with a few edges missing (by a couple of theorems)
trivalent graphs with a bridge $^\pilcrow$

but the general classification problem has thus far eluded the best efforts of Vizing and many others. There are interesting links here with polyhedral decompositions (aka cyclic double covers) and embeddings in surfaces.

A(n edge-)critical graph is a connected graph of class II but such that removing any of its edges makes it class I. As often in graph theory, such a minimality condition imposes a certain amount of structure on the graph. There are conjectures...

Almost all graphs are in class I

Let $\char93 G_n$ be the number of graphs with vertices, and $\char93 G^{\,\rm I}_n$ the number of them in class I.

Theorem (P. Erdős and R. J. Wilson) $\quad$

$\begin{displaymath} \lim_{n\to\infty} {\char93 G^{\,\rm I}_n \over \char93 G_n} = 1 \end{displaymath}$

So graphs of class II get relatively rarer for larger graph sizes. The absolute numbers do still increase. For cubic graphs for instance, we saw every one with a bridge is in class II. Bridgeless cubic graphs of class II are a bit thinner on the ground. By the four-color theorem, via Tait coloring, we know all of them are non-planar. Rarer still are those of them with girth at least five (and some non-triviality conditions); they are so hard to find that Martin Gardner dubbed them snarks. The Petersen graph is one, and a few infinite families of snarks have been found.

$\dagger\quad$ $\rho=2$ , so an edge- $\rho$ -coloring must use alternating colors. For odd that's impossible.

$\ddagger\quad$ $\rho=n-1$ , note it is also the valency of every vertex. Fix two colors and . A Kempe chain can only terminate at a vertex where one of those colors is missing but in we cannot afford to miss any color at any vertex, so every is a cycle of even length and together they visit all vertices. For odd that's impossible. For even there are ways to construct the coloring (try it).

$\pilcrow\;\;$ $\rho = 3$ and again the valency of every vertex. For the same reason as in the previous note, every or or is a cycle. Every edge must be in two such, but a bridge cannot be part of a cycle.

Bibliography

Ore67: OYSTEIN ORE, The Four-Color Problem,
Acad. Pr. 1967, ISBN0125281501
Long the standard work on its subject, but written before the theorem was proven. Has a wealth of other graph theory material, including proofs of (improvements of) Vizing's and Shannon's theorems.
FW77: S. FIORINI and R. J. WILSON, Edge-colourings of graphs, Pitman 1977, ISBN0273011294
The first ever book devoted to edge-colorings, including material previously found only in Russian language journal articles. Has proofs of Vizing's and Shannon's theorems.
SK77: THOMAS L. SAATY and PAUL C. KAINEN,
The Four-Color Problem: assaults and conquest,
McGraw-Hill 1977; repr. Dover 1986, ISBN0486650928
Wonderfully broad, not only focussing on the usual route to the Appel-Haken proof but also giving lots of other material.
Wil02: ROBERT A. WILSON, Graphs, Colourings and the Four-colour Theorem, Oxford Univ. Pr. 2002, ISBN0198510624 (pbk), http://www.maths.qmul.ac.uk/˜raw/graph.html (errata &c.)
A good general course in graph theory, with special focus on the four-color theorem and details of the Appel-Haken proof. Has proof of Vizing's theorem.

"Vizing's theorem" is owned by marijke.

(view preamble)