+ All Categories
Home > Documents > On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

Date post: 23-Jan-2016
Category:
Upload: ajoichu
View: 219 times
Download: 0 times
Share this document with a friend
Description:
By Vincent Edward
Popular Tags:
98
ON EDGE-CRITICAL GRAPHS AND THE NOTION OF VERTEX INDEPENDENCE IN GRAPHS by VINCENT EDWARD FA TICA B.S., Syracuse University, 1972 M.S., Syracuse University, 1977 M.Ph., Syracuse University, 1985 ABSTRACT OF DISSERTATION Respectfully submitted to the Graduate School of Syracuse University in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Mathematics August 1990
Transcript
Page 1: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

ON EDGE-CRITICAL GRAPHS AND THE NOTION

OF VERTEX INDEPENDENCE IN GRAPHS

by

VINCENT EDWARD FA TICA

B.S., Syracuse University, 1972

M.S., Syracuse University, 1977

M.Ph., Syracuse University, 1985

ABSTRACT OF DISSERTATION

Respectfully submitted to the Graduate School of Syracuse University in

partial fulfillment of the requirements for the degree

of Doctor of Philosophy in Mathematics

August 1990

Page 2: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

/

(

ABSTRACT

Chapter one provides an introduction to and an overview of the work. The overview is condensed in this abstract. For the convenience of the reader, the basic notions and terms of graph theory are presented in chapter two.

Chapter three gives an account of most of the major known results on graphs which are edge-crticial with respect to independence number. Chapter four presents three characterizations of these edge­critical graphs.

The following is shown in chapter five. For the task of decomposing a graph into disjoint subgraphs so that each inherits a maximum independent set of vertices from such a set in the given graph, so that the sum of their independence numbers equals that of the given graph, and so that none of them may be further decomposed, the edge-critical graphs are necessary and sufficient to serve as the decomposing subgraphs.

In chapter six it is observed that two of the Platonic graphs are not decomposable in the sense of chapter five. Decompositions of the other three are given, that of the dodecahedron being particularly striking. Chapter seven presents general constructions yielding connected edge-critical graphs.

In chapter eight the following is shown. For the task of covering a graph with subgraphs so that the sum of the independence numbers of the subgraphs equals that of the given graph, the edge­critical graphs are necessary and sufficient to serve as the covering subgraphs. Several conjectures regarding generalized covering theorems and edge-critical graphs are made.

Chapter nine gives numerous technical results aimed at proving one of the conjectures of chapter eight, and allowing for an elementary and graph theoretic proof of an extension, due to Gerards, of a celebrated theorem of Konig. Chapter ten presents this proof, and gives an account of the progress made in the attempt to prove the conjecture mentioned above. This conjecture, if verified, would further extend Gerards' result.

Chapter eleven gives a useful corollary to a theorem of Graver and Yackel on the independence number of a tree, and points out its relation to the preceding work.

Page 3: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

ON EDGE-CRITICAL GRAPHS AND THE NOTION

OF VERTEX INDEPENDENCE IN GRAPHS

by

VINCENT EDWARD FA Tl CA

B.S., Syracuse University, 1972

M.S., Syracuse University, 1977

M.Ph., Syracuse University, 1985

DISSERTATION

Respectfully submitted to the Graduate School of Syracuse University in

partial fulfillment of the requirements for the degree

of Doctor of Philosophy in Mathematics

August 1990

Approved --------------

Date

Page 4: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

( \

Copyright 1990

VINCENT EDWARD FA TICA

( ii

Page 5: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

To my parents

John Joseph Fatica

and

Jean Moore Fatica

and to my grandmother

Carolyn Bertha Moore

in memory of her

iii

Page 6: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

Contents

Chapter page

1. Introduction and Overview ....................................................................... 1

2. Graph Theory ......................................................................................... 5

3. The edge-critical graphs ........................................................................ 10

4. Characterizations of connected edge-critical graphs ................................ 16

5. Decompositions of Graphs ...................................................................... 20

6. The Platonic graphs .............................................................................. 24

7. Some methods for constructing edge-critical graphs ................................ 30

8. Coverings of Graphs ............................................................................. 42

9. Various Results .................................................................................... 49

10. Gerards' Extension of Konig's Theorem .................................................. 75

11. A note on a theorem of Graver and Yackel ............................................ 82

iv

Page 7: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

1

CHAPTER 1 - Introduction and Overview

The subject of this thesis is graph theory. For the benefit of the reader

who is unfamiliar with the basic notions and terms of the subject, these are

provided in the next chapter. For more in-depth presentations of the rudiments

of graph theory, the reader is referred to books by Berge ([B]), Bondy and

Murty ([BM]), Harary ([H2D, and Graver and Watkins ((GW]). The treatment

given in the last of these is particularly encyclopaedic and places great

emphasis on its mathematical rigor and its efficient use of notation and

symbolism. For the purposes of this introduction and overview, it is presumed

that the reader is familia.r with graph theory.

As mentioned above, chapter 2 presents the basic notions and terms of

graph theory. Chapter 3 introduces the notion of edge-critical graph. Edge­

critical graphs, per se, were studied for a brief time, by an apparently small

number of researchers, among them such notables as Lovasz and Gallai. The

collection of literature on edge-critical graphs is quite compact, and the

theorems presented in chapter 3 are indeed nearly all of the major results in

the area. Some of these are used to advantage later in this work. The others

are inlcuded in chapter 3 for their historical significance.

In chapter 4, three characterizations of edge-critical graphs are given.

None are to be found in the literature. Later in the work, two of these are put

to some use.

Chapter 5 asks and answers the following vaguely stated question: "Can a

graph be taken apart, or decomposed, into smaller, perhaps more manageable

Page 8: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

2

pieces, in such a way that the phenomenon of vertex independence in it may

somehow be understood in terms of the same phenomena in these smaller

pieces?" This question is made precise, and it is shown that the edge-critical

graphs are, in a very natural way, both necessary and sufficient to serve as

the "pieces".

In chapter 6, we will look at a. few familiar graphs in light of the

developments of chapter 5. We will consider the graphs formed by the

vertices and edges of the Platonic solids: the tetrahedron, the octahedron, the

cube, the dodecahedron, and the icosahedron, and take them apart (as much as

we can) in the manner of chapter 5. The view of the dodecahedron so obtained

is particularly striking.

In chapter 7 are presented constructions of edge-critical graphs. They

yield arbitrarily large graphs, with arbitrarily large deficiency (a parameter

according to whose value edge-critical graphs are classified), and which are

highly connected.

In chapter 8, we turn our attention to coverings of graphs, a notion

closely related to that of decompositions of graph. We ask: "Can we 'cover' a

graph by subgraphs in such a way as to gain information about vertex

independence in it from considering vertex independence in these subgraphs?"

The question is precisely posed, and it is shown that in a certain very natural

sense, the edge-critical graphs are necessary and sufficient to serve as the

covering subgraphs.

Page 9: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

3

While still in chapter 8, we introduce the reader to a celebrated theorem

of Konig and observe that it can be easily formulated in terms of edge-critical

graphs. Proposition 8.4 gives a "sweeping" extension of Konig's theorem which,

because its hypotheses are so strong, is, unfortunately, of little or no practical

value. However, several stronger (and quite reasonable) conjectures are made.

Chapter 9 has a two-fold purpose. We embark on an attempt at proving a

conjecture (9.2), which, though it is merely the first unresolved instance of a

more general conjecture from chapter 8, would, if it were determined to be true,

provide an extension of Konig's theorem which goes beyond those known to

date. These efforts unfortunately do not come to fruition but they do proceed

a long way in a direction which it is hoped the reader agrees is worthy of

continued investigation.

At the same time as we are trying to prove the above-mentioned

conjecture, however, we are accumulating results sufficient to allow us to give

an elementary proof of an extension of Konig's theorem due to A.M.S. Gerards.

Gerards' theorem is weaker than our conjecture 9.2 (indeed, it would be an

immediate consequence of 9.2), and his proof uses advanced notions and

techniques, most notably those of linear programming and matroid theory. The

elementary proof presented here will make the proof of this extension of

Konig's theorem accessible to a greater audience.

Chapter 11 may be considered an epilogue. It deals with a notion

communicated to this author privately by Jack Graver. It is precisely this

notion which proved to be the seed from which all of these researches grew.

Page 10: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

4

The material presented in chapter 11 is not without connection to the

earlier work (indeed we'll see a rather particular instance of the kind of

decomposition mentioned in chapter 5). Also, the main result of chapter lJ,

while merely a corollary to a theorem of Graver and Yackel, is interesting in

itself. Finally, it is hoped that the inclusion of this material will serve as a

small gesture of this author's appreciation for the inspiration, assistance, and

encouragement that Jack Graver has provided him.

Page 11: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

5

CHAPTER 2 - Graph Theory

We shall use the simple term "graph" to connote what should be more

properly termed "finite, simple, undirected graph without loops". For our

purposes, a ~ G consists in a finite set V(G) (V, for short), the elements of

which are called the vertices of G, paired with another set E(G) (E for short),

whose elements, called the edges of G, are doubleton subsets of V. The sets V

and E, aside from the stipulations that V be finite, and that each element of E

be a doubleton subset of V, are arbitrary. We shall often write G = (V, E) to

denote the graph G whose vertex set is the set V, and whose edge set is the

set E.

For a pair u E V, v E V of vertices of a graph G = (V, E), if e = {u, v} E

E, (i.e., e is an edge of G), we shall say that the vertices u and v are adjacent,

or that u is adjacent to v, and vice versa. We shall also say in this case, that

the vertex u (or v), is incident with, or is an endpoint of, the edge e; we shall

also say that u and v are neighbors. Edges sharing a common vertex are said

to be adjacent. The notion of graph itself and this language that we have

adopted, make it possible for us to put to very good use our geometric

intuition, for these objects can so easily be "pictured". As an example,

consider the graph K4 = (V, E) where V= {a, b, c, d} and E = {{a, b}, {a, c}, {a, d},

{b, c}, {b, d}, {c, d}}. (The name, K4, of this graph, a graph which will be highly

significant throughout this work, will be explained below.) K4 may be pictured

as i 11 Ill ustra ti on 1:

Page 12: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

6

Illustration 1: Two views of the graph K4 •

d b

In the Illustration, the heavy dots labelled a, b, c, and d represent the

vertices of K4 , and a line drawn between two dots reflects the fact that the

vertices represented by these dots comprise an edge of K4 •

K4 is just one example of a complete ~ that is, a graph in which

every possible pair of vertices comprises an edge. Other complete graphs (named

K11 K2 , etc. according to their numbers of vertices) are depicted in Illustration 2.

Illustration 2:

l l<:i_ Ks

Notice that in the case of K5 , that the lines drawn to represent the edges

were allowed to cross (their crossing is in fact of necessity, though a proof of

that fact is not within the scope of this work; the interested reader should

consult (CW]). It will often be necessary, or desirable, to do this. It is

understood that at the point where these lines cross, there is no vertex (there

not being a heavy dot there), and it should be imagined that the "edges" do not

in fact meet.

Page 13: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

7

The graphs Kn (n = 1, 2, .. .) are special in that their edge sets are

"complete". Another set of graphs is the collection of cvcles (or circuits), C3 ,

C4 , ••• (for so named obvious reasons), some of which are pictured in Illustration

3. The length of a cycle is the number of edges in it.

Illustration 3:

C3 . Cs

Before we can give an adequate definition of these graphs, we need to

consider a couple of new notions. By the valence (or degree) of a vertex of a

( graph is meant the number of edges incident with it; for a given vertex v of a

graph, its degree is denoted p(v). Furthermore, a graph is call connected, if for

every pair u and v of vertices of a graph G, there exists some sequence

u 0 , u1' .. ., Un of vertices of G for which u 0 = u, Un = v, and {uH ui+1} is an edge

of G for each i = O, 1, ... , n - 1. Such a sequence is called a path from u to v.

Geometrically speaking, a graph G is connected if it is possible to get from any

vertex to any other vertex by "travelling" along the edges of G.

All of the graphs mentioned and depicted so far are connected. The

cycles C3 , C 4 , • • • may now be easily defined as the collection of connected

graphs for which every vertex has degree 2. The graph G = ({a, b, c, d}, {{a,

b}, {c, d}}) depicted in Illustration 4 is an example of a graph which is not

connected. (

Page 14: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

8

Illustration 4: A graph which is not connected

0.. ··---------· b

c d Later in this work, we shall come across more refined notions of

connectedness. Presently, we need to explore a few more of the fundamental

notions and relations of graph theory.

As mentioned earlier, those vertices adjacent to a given vertex, v, are

called the neighbors of v; accordingly, the set of neighbors of v is called the

neighbor set of v and is denoted N(v).

A graph H = (V(H), E(H)) is said to be a subgraph of the graph

G = (V(G), E(G)) if (i) V(H) C V(G), and (ii) E(H) E(G) n P2(V(H)), where

P2(V(H)) denotes the collection of doubleton subsets of V(H). Certain types of

subgraphs of a given graph will be of particular importance. If G = (V, E) is a

graph and v E V, then G - v will denote the subgraph of G resulting from

deleting from V the vertex v, and deleting from E those edges incident (in G)

with v. Similarly, if G = (V, E) and S C V, then G - S will denote the

subgraph of G resulting from deleting from V the set S of vertices, and

deleting from E those edges incident with an element of S. And too, if G = (V,

E) and F C E, then G - F, (G - e if F = {e}), will denote the subgraph of G

resulting from the deletion from E of those edges in F.

If a graph G has as a subgraph a cycle C then by a chord of C is meant

Page 15: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

9

an edge of G not on C, but incident with two vertices of C.

In a graph G = (V, E), a set I C V is said to be independent if no two of

its members are adjacent. Such a set whose cardinality is a maximum among

the cardinalities of all independent sets of vertices of G is called a maximum

independent set (of vertices of G), and the cardinality of a maximum

independent set is called the independence number of G, and is denoted o:(G). In

keeping with this notation, and for the sake of conciseness, a maximum

independent set will hereinafter be called an a-set of G.

A few other numbers associated with a graph G CV, E) are of interest:

(i) v(G) = IVCG) I

(ii) e(G) IE(G) I, and

(iii) 6'(G) v(G) - 2o:.(G), (v - 2a. for short), which has by a few authors been

called the deficiency of G.

In [Gl], A. George has written:

"Finding a largest set of mutually non-adjacent ... points

[an a-set, to us] is one of the oldest problems of graph theory.

This problem arises in many different contexts (e.g. chessboard

problems, coding theory, network theory, matching problems, etc.)."

It is precisely this notion of independence to which the remainder of this

work is dedicated.

Page 16: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

10

CHAPTER 3 - The edge-critical graphs

In our investigation of vertex independence in graphs, we are asking a few

basic questions, which will be formulated more precisely later:

(i) Is it possible to take apart a given graph G in such a way that we may

gain information about the phenomenon of vertex independence in G by

studying the same phenomenon in each of the pieces?

and, if this is the case

(ii) Can we restrict our attention to any particular, hopefully small and well­

defined, class of graphs to serve as these "pieces"?

and, if this too is the case,

(iii) What can we say about this particular class of graphs?

Before attacking these questions, we need to be introduced to a particular

class of graphs, the edge-critical graphs.

Definition: A graph G

every edge e of G.

(V, E) is said to be edge-critical if cx(G - e) > cx(G) for

That is, G is edge-critical if the removal of any one of its edges causes

the independence number to increase.

The edge-critical graphs were first considered by Zykov ([Z]) in 1949, and

the list of those who have studied them since then is short, though it includes

some of the most prominent names in graph theory. And the body of literature

on edge-critical graphs is consequently small and compact.

Page 17: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

11

We will present below many (in fact, almost all) of the basic results on

edge-critical graphs. A few, mainly more technical, results will be saved for

later chapters. Fairly thorough accounts of the major results on edge-critical

graphs may be found in [G1J, [H2J, and [L2].

3.1 (see [B], p. 287) Let G = (V, E) be a connected edge-critical graph,

and let U V have the property that G - U is not connected (such a set U of

vertices is called a cut-set of G). Then the subgraph of G ~anned by the

vertices of U (i.e. G - (V - U)) is not a complete graph.

3.2 (due to Berge, see [BJ, p. 286) In an edge-critical graph, any pair of

adjacent edges lies on a chordless cycle of odd length.

3.3 (due to Hajnal, see [H1J) If G is an edge-critical graph without

isolated vertices (ie: without vertices of degree 0), then for every independent

set S of G, IN(S) I ~ IS J. [ N(S) = U N(V) ] vES

3.4 (ibid) If G is an edge-critical graph without isolated vertices, then

for every vertex v of G,

p(v) ~ v(G) - 2cx(G) + 1 = o(G) + 1.

3.5 (a corollary to 3.4, due to Erdos and Gallai, see [B], p. 293) An edge-

critical graph G contains at least 2cx(G) - v(G) isolated vertices.

As the reader may be coming to realize, the parameter o(G) = v(G) - 2cx(G)

is of great importance in classifying the edge-critical graphs. We shall adopt

Page 18: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

the following notation:

r = the collection of all edge-critical graphs

r n = the collection of edge-critical graphs G for which

o(G) = v(G) - 2a.(G) = n.

A = the collection of connected edge-critical graphs

l:ln = A n r n· (n = 0, 1, 2, ... ).

12

The fact that a connected graph has no isolated vertices, together with 3.5

above provide the range of values of n in the definition of An.

For convenience, we will let ;:1_ 1 denote the collection of graphs whose

only member is the graph K10 the graph with one vertex and no edges. (This

graph is edge-critical and connected, both trivially, and indeed o(K 1) = -1).

The next three results are of particular importance in the remainder of

this work.

3.6 (due to Andrasfai, see [A], p.10) A0 is the collection whose only

member is K2, the graph with two vertices and one edge.

3.7 (ibid) A 1 = {C3 , C5 , C7, ... } i.e., the collection of cycles of odd length.

A graph is called an odd subdivision of K4 or, to us, a very-odd-K4 if it

can be obtained from the graph K4 by replacing some or all of the six edges of

K4 by mutually disjoint paths of odd length. Illustration 5 depicts some very­

odd-K4's.

Page 19: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

13

Illustration 5:

3.8 (ibid) t.. 2 is the collection of graphs consisting solely of the very-odd-

K/s.

Though it is not the last in our list, investigation into the edge-critical

graphs seems to have come to an end with the following finite basis theorem.

3.9 (due to Lovasz, see [Ll], p. 726) For each of 6 = 1, 2, 3, ... there

exists a finite collection of graphs in A 8 with the property that all the graphs

in t..0 can be obtained from these by the replacement of some or all of their

edges by mutually disjoint paths of odd length.

Note: We've seen in 3.7 that the single graph K3 provides such a "finite basis"

for A1, and in 3.8 that the single graph K4 provides one for .6. 2 • Such

simplicity however does not remain the rule. It is known that for 6 > 3 the

number of graphs needed is greater than one. The constructions given later in

chapter 7 provide two such examples for each 6 > 3.

Lovasz finishes the proof of 3.9 by deducing that a graph in the "basis" 8

for a given o may have no more than 22 vertices. This bound is not believed

to be sharp. In the case 6 = 3, this bound is approximately 1.84 X 1019, In

(S3], Suranyi gives a better bound, namely 57.

Page 20: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

14

We conclude this chapter with two theorems which will be valuable to us

later.

As 3.8 and 3.9 suggest, this process of odd-subdivision (and its "reversal")

preserve the property of edge-criticality. This fact is stated formally in the

following theorem, due to Andrasfai:

3.10 (see [A], p. 14)

a) If G = (V, E) E. f and v E. V with p(v) = 2, v having neighbors u and

w, then the graph G' obtained from G by the deletion of the vertex

v and the edges {u, v} and {v, w}, and the subsequent identification

of the vertices u and w, is also in f; moreover, o(G') = o(G).

b) If G = (V, E) E. r, v E. V, and p(v) > 2, and if v is split into two new ver-

tices v 1 and v 2 , while N(v) is partitioned into the non-empty sets N(v 1)

and N(v2 ), and if a new vertex u, and the edges {u, v 1} and {u, v 2} are

added, then the resulting graph, G', is also inf; moreover o(G') = o(G).

Before we can appreciate the following beautiful theorem of Gallai,

characterizing the edge-critical graphs which are connected, but not highly

connected, we must elaborate on the notion of connectedness itself.

Definition: A connected graph, G = (V, E), is said to be k-connected

(k = 1, 2, .. .) if for every S C V with ISi :;;; k - 1, G - S is also connected.

That is, G is k-connected if G has no cut-set containing fewer than k

vertices. By the connectivity of a graph, G, we mean the greatest integer k

for which G is k-connected.

Page 21: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

15

Gallai's theorem, below, characterizes the edge-critical graphs of

connectivity 2 (i.e., the (2 but not 3)-connected ones). Note that 3.1 implies

that connected edge-critical graphs are 2-connected.

3.11 (due to Gallai, comminicated privately to Andrasfai, see [A], p. 13)

a) Let G1 and G2 be two graphs in A, having no vertices in common, and

each having more than one edge. Let e = {uJJ vJ be an edge of G1

and let x be a vertex of G2 • Let Gi be the graph G1 - e and let G2

be obtained from G2 by splitting x into two vertices u 2 and v 2 while

partitioning N(x) into the non-empty N(u 2 ) and N(v 2). Let G be the

graph obtained by identifying u 1 and Vu of G~ with u 2 and v 2 of G2

(respectively). Then G E A.

b) Let G E A and suppose {u, v} is a cut set of G. Then {u, v}

is not an edge of G (cf., 3.1) and the graph G - {u, v} has exactly two

components. Let N~' and N2' be the sets of vertices of these

components, and for i = 1, 2, let Ni = Ni' U {u, v}. Let Gi be the

subgraph of G spanned by N~ . Then exactly one of the graphs Gi has

the property that the graph Gi obtained from it by adding the edge

{u, v} belongs to A; and the other one (and only the other one) has

the property that the graph, Gi, obtained from it by identifying u and

v belongs to A.

c) In both a) and b)

v(G) = v(G 1) + v(G 2 ) - 1 and

cx.(G) = cx.(G 1) + cx.(G 2 ), and consequently

o(G) = 0CG1) + 0CG2) - 1.

Page 22: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

16

CHAPTER 4 - Characterizations of connected edge critical graphs.

No characterizations of edge-critical graphs appear in the literature.

Three are given below. The two characterizations contained in 4.1 characterize

edge-critical graphs in terms of the local interaction between a-sets and some

particular sets of vertices. The third characterization, given in 4.2 [as will be

seen later] characterizes edge-critical graphs by a property which will lead

quite naturally to one of the next chapter's theorems on decompositions of

graphs in general.

4.1 - For a connected graph G, the following are equivalent.

(a) G E A;

··-·- tlr)- for every vertex v oT G, N(vJ is minimal wTffi r-espect To set indusion

among the subsets S of V(G) with the property that S n I ~ 0 for

every O'.-set I of G - v;

(c) for every vertex v of G, and for each vertex x E N(v), there exists

an o.-set I of G with the property that I n N(v) = {x}.

Proof: (a implies b) Suppose G E A and let v be a vertex of G. If N(v) = 0,

then, being connected, G is the graph K1' whose only vertex is v; in this case

(b) follows trivially. Suppose N(v) ~ 0 and let u E N(v) and let e = {u, v}.

Let I be an O'.-set of G - e. II I > a(G) since G is edge-critical; in fact:

(i) contains both u and v, for if not, I would be independent

in G, which is absurd. I - {v}, being independent in G, is

consequently an a.-set of G and thus cx(G - v) = cx(G).

Page 23: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(ii) Every o:-set of G - v meets N(v), for if one, say J, did not, J U {v}

would be an independent set in G with

IJ u {v}I !JI + 1 = cx(G - v) + 1 = <x(G) + 1

which is absurd. And,

17

(iii) N(v) has the desired minimality. To see this suppose S is a proper

subset of N(v) and let w E N(v) - S. Let e denote the edge {v, w} of

G. Let I be an a.-set of G - e. II I = cx(G) + 1 (as in (i)). Moreover,

I ll S = 0 since v E I (as in (i)). Thus, I - {v}, which has cardinality

a(G) = a(G - v) is an a-set of G - v which fails to meet S.

(b implies c) Suppose condition (b) holds and let v be a vertex of G. If

N(v) = 0, then (c) is vacuously true. So suppose w E N(v). Let e denote the

edge {v, w} of G and let I be an ex-set of G - e. I n N(v) = {w}, for I is

independent and contains v. This is as desired.

Cc implies a) Suppose condition (c) holds. Let e = {v, w} be any edge of

G. Let I be an a.-set of G with I n N(v) = {w}. I U {v} is independent in G - e

and thus

a(G - e) :;:;:; II U {v} I II I + 1 = a(G) + 1

implying G E t:... 0

The following definition will facilitate the statement and proof of our

third characterization of connected edge-critical graphs.

Page 24: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

(

18

Definition: A subgraph H of a graph G is called hereditary if for every o:.-set

I of G, I n V(H) is an o:.-set of H.

4.2 A connected graph is edge-critical if and only if it has no proper

hereditary subgraphs.

Proof Suppose a connected graph G is edge-critical and suppose the proper

subgraph H of G is hereditary.

Case 1: If V(H) = V(G), then o:.(H) = o:.(G) for if I is an a-set of G,

I n V{H) = I. Now, H being a proper subgraph of G, there must be an edge e of

G which is not an edge of H. Let I be an a-set of G - e. Since G is edge­

critical, III = a(G) + 1 which is the same as a(H) + 1. But I is independent in

H. This is absurd.

Case ~: If V(H) ~ V(G), let K be the subgraph of G spanned by the vertices in

V(G) - V(H). Since H is hereditary, for every O'.-set I of G, II n V(H) I = ryJH)

and so for every a-set, I, of G, II n V(K)I = a(G) - a(H).

Let e be an edge connecting a vertex v of H to a vertex w of K and let

be an a-set of G - e. Since G is edge-critical, II I = cx(G) + 1. Certainly

cannot contain more than a(H) vertices of H. On the other hand, if II n V(H) I

::;; cx(H), then I - { v} is an ex-set of G with

!CI - {v}) n V(H)I = II n V(H)I - 1 ::: cx(H) - 1

contrary to the supposition that H is hereditary.

Page 25: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

19

On the other hand, suppose G is connected and not edge-critical. Let e be

an edge of G with the property that a(G - e) = o(G). Then, G - e is itself a

proper hereditary subgraph of G. 0

Page 26: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

20

CHAPTER 5 - Decompositions of Graphs

We now turn our attention to the questions posed at the beginning of

chapter 3. We shall formulate the first two of them more precisely, and

answer them.

Definition: Let G = (V, E) be a graph. The subset Va.CG) of V is defined to be

the union of all the a_-sets of G. Va.CG) is called the a_-support of G.

Definition: By an o:.-decomposition of a graph G, we mean a collection

{Hu Hz, ... , Hk} of subgraphs of G with the following properties:

(a) each H, is a hereditary subgraph of G,

k (c) U VCH;) = Va.CG).

i=l

Note: Of great importance is that the properties (a), (b), (c) above imply that if

k {Hu Hz, ... , Hk} is an cx.-decomposition of G, then a(G) = 2:cx.(Hi).

i=l

Definition: By the trivial a-decomposition of G, we mean the collection {G}.

Definition: An a-decomposition of a graph G will be called a preferred

a-decomposition if none of. the Hi have non-trivial a-decompositions.

Note: If {H1, Hz, ... , Hk} is a preferred ex-decomposition, then each H; is

connected.

Page 27: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

21

5.1 If G is connected and edge-critical, then G has no o.-decomposition other

than the trivial one.

Proof: 4.2 characterizes connected edge-critical graphs as those connected

graphs having no proper hereditary subgraphs. From this alone, the desired

result follows. D

5.1 tells us that if we desire a collection of graphs, the members of which

suffice to serve as the "building blocks" for the kind of decomposition

mentioned above (i.e., to serve as the decomposing subgraphs), then we are

obliged to include in that collection the connected edge-critical graphs. That

is, that they are necessary. We shall soon see that they are sufficient as well.

5.2 If G is not edge-critical, then G has a non-trivial ex-decomposition. (Note

that only when G is connected and V O'JG) = V(G) is {G} an ex-decomposition.)

Proof: First note that if the subgraph G0 of G spanned by the vertices in

VaCG) is edge-critical, then the components of G0 constitute a (preferred) ex­

decomposi ti on.

If G0 has no edges, then G0 is a collection of isolated vertices and so is

edge-critical, a case which we have already considered.

So let e0 be an edge of G0 with the property that ex(G0 - e0 ) = ex(G0 ), and

let G1 = G0 - e0 • Continue in the following way: having constructed Gi, with

V(Gi) = V(G 0 ) and ex(Gi) = ex(G 0 ), if Gi is not edge-critical, let ei be an edge of

Gi with the property that o:.(G, - ei) = ex(Gi) and let GH 1 = Gi - e 1• This cannot

Page 28: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

continue ad infinitum. Eventually, say when i = N, GN will be edge-critical,

with V(GN) = V(G 0 ) and o:(GN) = ex(G). At this point, the components of GN

(which are each edge-critical) constitute a (preferred) ex-decomposition of G. D

5.3 (Lemma) If H is connected and edge-critical, then there exists no

collection H1, ••• , Hk of disjoint, connected subgraphs of H, the union of whose

vertex sets span H, and which has the property that o:.(H) = 2:)x(Hi).

Proof: Let H be connected and edge-critical and suppose that such a collection

{H1' ... , Hk} does exist. Let e be an edge of H joining vertices in different H1•

Since H is edge-critical, ex(H - e) = ex(H) + 1. Let I be an o:.-set of H - e. Then

I is independent in the subgraph of H whose components are the Hi. But

III = ex(H) + 1 = 2:exCH) + 1, which is absurd. D

( \

The following is almost immediate:

5.4 If a graph G has an ex-decomposition whose elements are connected edge-

critical subgraphs, this ex-decomposition is a preferred ex-decomposition.

Proof: If {H1, ••• , H1G} were an a-decomposition of G with each H1 connected and

edge-critical, while not being a preferred ex-decomposition, this would imply the

existence, for at least one of the Hit of a collection of subgraphs, the existence

of which is precluded by 5.3. 0

5.5 Every graph G has a preferred ex-decomposition whose elements are edge-

critical subgraphs of G.

Page 29: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

23

Proof: By 5.1 and 5.4, if G is edge-critical, the components of G, which are

also edge-critical, comprise a preferred a-decomposition of G.

If G is not edge-critical, then the construction used in the proof of 5.2

yields an a-decomposition whose elements are connected edge-critical subgraphs.

By 5.4, this is a preferred a-decomposition. 0

This shows that the connected edge-critical graphs are sufficient for the

task of decomposing graphs in the sense of what we've called a preferred

cx-decomposi ti on.

Page 30: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

24

CHAPTER 6 - The Platonic graphs

The Platonic solids have been of interest for a very long time, and more

recently, their vertices and edges have become interesting examples of graphs.

We digress here to view these well known graphs in light of the foregoing

remarks. (Note that for each of these graphs, Va(G) = V(G).)

1. The tetrahedron, depicted in Illustration 6, is an edge-critical graph.

Indeed, it is the graph K4, the scle building block of the class .6. 2 of

connected edge-critical graphs.

Illustration 6: Tetrahedron (K4 )

2. lllustration 7 depicts the octahedron.

Page 31: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

Illustration 7: Octahedron

I

I \ I

25

In the illustration, all the edges, both solid and broken, are edges of the

octahedron. The two subgraphs (triangles, three-cycles) show a preferred

a-decomposition of the octahedron into two edge-critical subgraphs, each with

6 = 1. The decomposition is unique up to automorphism.

3. Illustration 8 depicts the cube (solid and broken edges) together with a

preferred a-decomposition (solid edges only) into 4 edge-critical subgraphs,

each with o = 0.

Page 32: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

26

Illustration 8: The cube

f': //1 ' I "

/ I /

I " /

"- /

~ ( V=- 8

I o\ =- 4 I d = 0

;b ~ / ' " / " I / " I

/ " k ~

( 4. The icosahedron, shown in Illustration 9, is, as will be shown, an edge-

critical graph.

Page 33: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

Illustration 9: The icosahedron

v:: /,;),_

o<. :. 3

d =- b

27

Let us first see that for the icosahedron, ex = 3. Referring to the

labelling in Illustration 9, if an independent set of vertices

(i) contains 0 or 1 of the x's, it may contain at most 1 of the v's and at

most 1 of the y's;

(ii) contains 2 of the x's, then, of the v's and y's, it may contain none of

one and at most 1 of the other;

(iii) contains 3 of the x's, it may contain none of the v's and none of the y's.

In any case, the cardinality of an independent set of vertices in the icosahedron

is limited to 3, and since {x2 , x4, x6 } is, in fact, independent, ex = 3.

In showing that the icosahedron is edge-critical, we need only check the

Page 34: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

28

removal of the edge {xl> x2}, for, by virtue of the symmetry of the icosahedron,

this edge may represent any of its edges. Upon removing {x1, x2 ), the following

set of four vertices is independent: {x 11 x2 , v 2 , y 2}.

Thus, the icosahedron is edge-critical. Moreover, o v - 2<X 12 - 2(3) =

6.

5. The most striking of these examples is the dodecahedron, depicted in

Illustrations 10 and 11. Illustration 10 gives a typical depiction of the

dodecahedron while Illustration 11 shows a preferred o:.-decomposition of it

into two edge-critical subgraphs, each with o = 2. (Recall that ~2 is the

collection of odd-subdivisions of K4 , or as we have called .them, the very­

odd-K4's.) The labellings of the vertices in the two illustrations are

consistent.

Illustration 10: The dodecahedron

V = .2..0 <:>(. :::. 8'

d = t

Page 35: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

29

Ill ustra ti on 11: Pref erred o:-decomposi ti on of the dodecahedron

- - - - - -.- -- -_., ........

/

/ ' D - - - - ' - - '+ '

/

~ ' I '\

\ F ~ b

B ;;...

G- ... 7 ... \

\ 10 \

' l --' - - - /'

/

' '>/'

' _,. -..----- - - - :- - -

With this view of the dodecahedron, it is quite easy to demonstrate that

its independence number is 8. Each of the decomposing very-odd-K/s has o: = 4

(having v = 10 and o = 2). Thus for the dodecahedron, o: ~ 8. Since

{A, C, F, I, 1, 4, 7, 10} is indeed of cardinality 8 and independent in the

dodecahedron, its independence number is equal to 8.

(

Page 36: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

30

CHAPTER 7 - Some methods for constructing edge-critical graphs.

Before continuing along more abstract lines, we will give a few

constructions which yield connected edge-critical graphs.

7.1 V(G) be minimal with respect to

set inclusion among the subsets of V(G) which have non-empty intersection with

every a-set of G. Let G' be the graph whose vertex set is

V(G') V(G) U {v},

where v is a "new" vertex, and whose edge set is

(i.e., add the new vertex v to G, and connect it with edges to each vertex in

U). Then,

G' E ll. and o(G') o(G) + 1.

Proof: cx(G') = cx(G) because U has non-empty intersection with every a-set

of G. Since G' is connected, we need only to show that the removal of any

edge from G' causes its independence number to increase.

If e is one of the edges of G, then o:.(G' - e) > o:.(G'), since G itself is

edge-critical. Now let e be one of the new edges, say, without loss of

generality, {v, v 1}. To conclude that a(G' - e) > a(G'), it will suffice to show

that G has an ex-set I with I n U = {v1}, for then I U {v} will be independent in

Page 37: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

31

G' - e and of cardinality o:.(G) + 1 > o:.(G').

(The argument that follows is similar to the one presented in the proof of 4.1.)

Some a.-set of G contains v 11 for if not, the set U - {vJ would contradict

the minimality of U. And too, if every o:.-set of G which contains v 1 also

contains some other element of U, the set U - {v 1} would again contradict the

minimality of U. Thus G has an o:.-set meeting U in precisely {v 1}. This is as

desired, and the proof is complete. D

Sets such as the set U in the preceding theorem do not in general make

themselves obvious.

corollary to 7 .1.

One type, however, does as is seen in the following

7 .2 (Corollary to 7 .1) If G is edge-critical and connected and if v E V(G) then

the graph G' obt ined from G by the addition of a new vertex u, and new edges

from u to each of the members of {v} U N(v) is also connected and edge­

critical.

Proof: The characterization of connected edge-critical graphs given in 4.l(b)

guarantees that the set {v} U N(v) has the necessary properties to serve as the

set U in the construction of 7 .1. D

Illustration 12 depicts how the construction of 7 .2 works in the particular

case of the complete graphs K1, K2 , K3 , •.• which are respectively elements of

A-1' A 0 , A 1, ••.•

Page 38: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

Illustration 12: The construction of 7 .2.

0

1<.:i.

+

+

+

> •

Another example is given in Illustration 13.

Illustration 13:

+

7 .3 Let G have vertices v 11 v 2 , ... , v Zk+H y 1, ... , y k+H x, and let

32

0 • k~

i> }(3

l<.'f

(i) the vertices v 1, v 2 , ... , v 21,+1 be the vertices of a chordless cycle in

G of length 2k + 1,

(ii) Yi have an edge to v 2 i-l and v 2 i (i 1, ... , k),

(iii) Yk+l have an edge to v 2k+H

(iv) x have an edge to each Yi·

Page 39: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

Then G c:: .C.k+i-

For k 3, the graph G is depicted in Illustration 14a.

Illustration 14:

v, Proof: v(G) = (2k + l) + (k + 1) + 1

Cb)

·v7 (x.)

3k + 3. o:.(G) = k + 1, for

33

(a) an ind.:::pendent set of G containing none of the Yi must contain x, and may

contain at most k of the vi, thus limiting its cardinality to k + 1,

(b) an independent set of G containing, say, j > 0 of the Yi may not contain x,

and may contain at most (k + 1) - j of the vi> again limiting its cardinality

to k + 1, and

(c) there is an independent set in G of cardinality k + 1, namely

6(G) = v(G) - 2o:.(G) = 3k + 3 - 2(k + 1) = k + 1.

To see that G is edge-critical, observe that

(a) if any edge {vi. v J} is removed, then an independent set of cardinality

k + 2 exists, containing k + 1 of the vi and containing x;

(b) if any edge {vi, y) is removed, then there exists an independent set of

cardinality k + 2 containing all the y /s and vi; and

Page 40: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

(c) if any edge {x, y ;} is removed, there exists an independent set in G of

cardinality k + 2 containing k of the Vi (but no neighbor of y J), together

with YJ and x. D

34

The graphs resulting from this construction are (2 but not 3)-connected;

indeed, the vertex Yk+i has degree 2. Andrasfai's construction (3.10) could then

be used to delete Yk+i and identify its two neighbors. The graphs which result

(the case k = 3 is depicted in Illustration 14 (b)), while they are not more

highly connected, are, since they have no vertices of degree 2, necessarily

members of Gallai's "finite bases" for the various Ak.

7.4 Let G have vertices vl> v 2 , ••• , v 2k+H xi> x2 , ... , x4k+Z• (k ::: 1). Let the vi

be the vertices of a complete graph K2 k+l and let the x1 be the vertices of a

chordless cycle of length 4k + 2. Finally, let each v 1 have as neighbors (other

than the other v's), the vertices x2 i and XzHZk+l (the subscripted indices being

considered modulo 4k + 2). Then G E A2k+P

Illustration 15 depicts the graphs G for k 1 and k 2.

Page 41: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

35

Illustration 15:

-x.,

'X-7

k::- I d =3 ) f<=;L cf =-S

\

Proof of 7.4: The set {x21 : i = 1, 2, ... , 2k + 1} is independent in G. Thus

o:(G) :?: 2k + 1. An independent set in G may contain no more than one of the

v's, and if such a set contains one of the v's, then it may contain only 2k of

the x's (since the inclusion of one of the v's precludes the inclusion of an

antipodal pair of x's). Thus cx(G) = 2k + 1. Since v(G) = (2k + 1) + (4k + 2) =

6k + 3, o(G) = (6k + 3) - 2(2k + 1) = 2k + 1.

G is certainly connected. To see that G is edge-critical, observe that

(a) if an edge between two of the x's is removed, say, without loss of

generality, {xll x4k+2}, then there exists an independent set of cardinality

2k + 2 containing 2k + 1 of the x's and one of the v's. As we have

supposed without loss of generality that {xll x 4k+2} was removed, this

independent set could be taken to be

Page 42: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

36

for 2k + 1 of the x's appear, but not x2 and x2k+3 , the x's which neighbor v 1•

(b) If one of the edges {v1, v) is removed, we begin constructing an

independent set I of cardinality 2k + 2 by including in it v 1 and v J• The

inclusion of vi in I precludes the inclusion of an antipodal pair of the x's,

leaving as candidates for inclusion in I two strings of x's of even length

(length 2k to be precise). In any string of vertices of even length 2k, an

independent set of k of them may be chosen so as not to include any

given one of them. If we now include in I an independent set of k

vertices in each of these strings which fails to include the neighbors of

vJ, we have arrived at the desired independent set in G - {v;, v.1} of

cardinality 2k + 2.

(c) If one of the edges {xH v J} is removed from G we may find the desired

independent set of cardinality 2k + 2 by simply choosing vJ along with the

2k + 1 vertices xH XHz• xH4 , ••• , xH'l,k (indices modulo 4k + 2), noting that

the other neighbor of v J among the x's (which is XHzk+r) is not chosen. D

One may consider the icosahedron (see chapter 4) to be constructed in the

following way: Let G have vertices v1' v 2 , v 3 , xll x 2 , ... , x6 , YH y 2 , y 3 • Let the

v's (respectively, the x's, the y's) be the vertices of a triangle (respectively, a

chordless 6-cycle, a triangle). For i = 1, 2, 3, let Yi also have as neighbors

Xzi-1' Xw and XzHl• and for i = 1, 2, 3, let Yi also have as neighbors Xzi• Xzi+H

and x 2H 2 (the subscripted indices on the x's being considered modulo 6). This

is the icosahedron (pictured below).

Page 43: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

37

Illustration 16:

V =I~

d ::. b

If we use the same method to attach two 4-cycles to an 8-cycle, we again

arrive at an edge-critical graph (see 7.5 below). This graph was also considered

by Watkins (see (W], p. 245) but for quite different reasons. The construction

in general, however, does not yield edge-critical graphs.

(respectively, the x's, the y's) be the vertices of a chordless 4-cycle

(respectively, 8-cycle, 4-cycle). Let v, also have as neighbors x 2i-1' Xw and

x2 H 11 and let y 1 also have as neighbors xw x2 Hi• and x2 H 2 (the indices on the

x's beng considered modulo 8). Then G E A8 •

Page 44: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

(

38

Illustration 17:

Proof: We will first show that ex = 4. There are three cases to consider,

according to whether a given maximum independent set I of vertices in G

contains 0, 1, or 2 of the y's.

Case 1: If I contains none of the y's, then either

(a) I contains none of the v's, in which case I contains at most 4 of the

x's,

(b) I contains 1 of the v's, in which case I contains at most 3 of the x's,

or

(c) I contains 2 of the v's, in which case I contains at most 2 of the x's.

Case 2: If I contains 1 of the y's, then either

(a) I contains 3 of the x's and none of v's,

(b) I contains 2 of the x's and at most 1 of the v's, or

(c) I contains 1 of the x's and 2 of the v's.

Page 45: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

(

Case 3: If I contains 2 of the y's, then either

(a) I contains 2 of the x's and none of the v's,

(b) I contains 1 of the x's and 1 of the v's, or

(c) I contains none of the x's and 2 of the v's.

39

In any case, the cardinality of I is limited to 4 and since {x 11 x3 , x5 , x7 } is

indeed independent, a. = 4.

In showing that G is edge-critical, we need to check the removal of only a

few of the edges, for the remainder of the cases follow from the high degree of

symmetry of G. The representative edges are:

(1) {y11 y 2}, representing a typical edge on either of the 4-cycles;

(2) {xl> x 2}, representing a typical edge on the 8-cycle;

(3)

(4) {yl> x 3}, representing typical edges from one of the 4-cycles to the 8-cycle.

The independent sets of cardinality 5 which arise upon the removal of

these representative edges are listed below, respectively:

(1) {yl> Y2, V11 V3, X3},

(2) {Xl> X2 1 X4, V3, y3},

(3) {y 1, x 2 , x6 , v 2 , v 4} and

(4) {y1' X3, X11 X5, y3}.

Lastly, o(G) v(G) - 20'..(G) 16 - 2(4) 8. D

The following very similar construction can be generalized, and yields

4-connected edge-critical graphs in A 4 , A 6 , A 8 , ....

Page 46: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

40

7 .6 Let G have vertices x 1, x2 , .. ., x2k+i• y 1, y 2 , .. ., Yn+I> (k 2: 1). Let the x's

and the y's be each the vertices of a chordless odd cycle of length 2k + 1, and

let X; also have as neighbors y,_ 1, Yi, and Yi+i (subscripts modulo 2k + 1). Then

G E A2k+2·

[In 7 .6, letting k = 1 result$ in the complete graph K6 , which is in A 4 •

The results for k = 2 and k = 3 are depicted below.]

~I

k=::i. ) d = 6 j <>

1~ = 3 )

Proof of 7.~: Clearly, v(G) = 4k + 2. Now consider the edges {x1, y 1}, {x2, y 2},

.. ., {x2k+H Yzk+ 1}. For the same reason that a collection of k + 1 vertices on a

2k + 1 cycle must contain an adjacent pair, a collection of k + 1 vertices of G

must contain a representative of each of some "consecutive" pair of these

edges, say {xH y 1} and {xi+l> yi+ 1}. But the vertices comprising these two edges

are pairwise adjacent, implying that a collection of k + 1 vertices of G cannot

be independent. Thus a.(G) :::;; k. And since {x2, x 4, ... , x2 k} is independent in G,

Page 47: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

41

and has cardinality k, a(G) k.

Therefore,

o(G) = v(G) - 2o:.(G) (4k + 2) - 2k 2k + 2.

The high degree of symmetry of G allows us to check only a few

representative edges of G in determining that G is edge-critical. They are:

(1) {x 1, x2}, representing an edge on either of the 2k + 1-cycles;

(2) {xu y 1} and

(3) {x1, y 2 }, representing an edge between the two 2k + 1-cycles.

The removal from G of each of these results in an independent set of k +

1 vertices. These independent sets are given, respectively, below.

(1) {x1, Xz, X4, X5, ... , X21J,

(2) {xll YH X3, X5, ... , Xn:-1},

(3) {x1' y 2• X4, X5, X3, .. ., Xzd.

So G is edge-critical, and, as desired, in .l2k+2· D

Page 48: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

42

CHAPTER 8 - Coverings of Graphs.

The notion of a-decomposition of an arbitrary graph G, presented in

chapter 5, while abstractly appealing, has a certain shortcoming: it presupposes

some knowledge of the phenomenon of vertex independence in G, at the very

least, knowledge about the set V o.(G). The question of determining the set

V o.(G) will not be considered in this work, and it seems to be, in general, a

difficult one.

So we will consider a closely related notion, that of a covering of a given

graph G by subgraphs, in the hope that we may understand vertex independence

in G, with only a knowledge of some properties of G which are relatively

explicit (compared to some knowledge of V O'.(G)) in the definition of G itself.

Definition. By a covering of a graph G, we mean a collection {H 1, H2 , •.• , Hk} of

k connected subgraphs of G with the property that V(G) = U V(Hi).

i=l

k

For any covering of G, o:.(G) ~ L:a(Hi), for if I is an a-set of G, i=l

k k k ex.CG) = II I I .u o n vmi) I ~ L: II n VCH;) I ~ :Lex.CH;).

t=l i=l i=l

Definition. By an ex-covering of a graph G, we mean a covering of G with the k

additional properties that cx(G) = L:o:.CH;) and that the H; are vertex disjoint. i=l

1' That is, a covering by disjoint subgraphs for which L:o:.CH;) is a minimum. It

i=l

will be seen that with some knowledge of the kinds of subgraphs which G has,

we can put very definite limits on the subgraphs needed to serve in an

Page 49: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

43

o:-covering.

The notions of O'.-decomposition and ex-covering are very similar. there

are two major differences:

(i) a-decompositions ignore vertices not in V o:(G), and

(ii) ex-coverings, while they do consist of hereditary subgraphs, do not

necessarily consist of minimal ones, as do ex-decompositions.

These differences are exemplified by the following simple ex.:rnple,

showing a graph ({a)), its ex-decomposition ((b)), and two ex-coverings ((c) and (d))

by edge-critical subgraphs:

(a..)

1 w

«..

j • • (C)

:i.. ~

• w

. . ~ (b)

• w

~ :;:. (d)

Note that the ve.:-tex x, which is not in Va does not appear in the ex.-

de~omposition. Note also that the edge {w, x} in (c), and the triangle in (d),

though they are hereditary subgraphs, are not minimal ones.

Page 50: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

44

8.1 Every graph has an a.-covering whose elements are in A.

Proof: Let G be a graph. If G is edge-critical then the desired result holds

trivially. If G is not edge-critical, then, as in the proof of 5.2, the removal of

edges without increasing oJG) until this can no longer be done yields a subgraph

of G whose components, which are each edge-critical, constitute an a-covering

of G. D

8.::. A connected edge-critical graph G has no a-covering other than the

trivial one, {G}.

Proof: Since, for an edge-critical graph G Va(G) = V(G), any a-covering of G is

also an a-decomposition of G. In 5.1, we have shown that a connected edge-

critical graph has no ex-decomposition other than the trivial one. The desired I

( result is immediate. D

8.1 and 8.2 show, respectively, the sufficiency and the necessity of the

connected edge-critical graphs for the task of providing coverings of graphs

which respect vertex independence in the sense of our a.-coverings. It is

unfortunate indeed that this class is so large, and that so little is known about

it. In light of this, one might hope that for a given graph, it would be possible

to place some limits on which connected edge-critical graphs are needed. The

first theorem along these lines, a celebrated theorem of Konig (see [BJ, [BMJ,

[H2], or [GWJ) well predates any consideration of edge-critical graphs or of

coverings in general which respect vertex independence in the sense of our a.-

coverings.

( We shall state Konig's theorem in a way in which it is often stated, and

Page 51: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

45

( then restate it in terms of the notions of this work.

Definition: A graph G is said to be bipartite if there exists a partition of V(G)

into sets V 1 and V 2 so that every edge of G has one of its vertices in V 1 and

the other in V 2 •

A common version of Konig's theorem may now be stated.

Konig's Theorem: Let G be a bipartite graph without isolated vertices. Let J&

be a covering of G [in our sense of covering] whose elements are each an edge

of G. If, among all such coverings of G, 1%1 is minimum, then

a_(G) 1%!. ( '

If, in Konig's theorem, we allow vertices as well as edges to serve as the

covering subgraphs, then we may insist that the covering subgraphs be disjoint,

for a pair of edges sharing a vertes, say {u, v} and {v, w}, may be replaced in a

covering by the edge {u, v} and the vertex w.

It is a well known and elementary fact that a graph is bipartite if and

only if it contains no odd cycles.

Putting these last two observations together, and noticing that the first

of them allows us to handle graphs with isolated vertices, we may restate

Konig's theorem as

Konig's Theorem: If G contains no odd cycles then G has an a-covering by

Page 52: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

vertices and edges;

or, as

8.3. Konig's Theorem: If G contains no subgraph in L}.u then G has an

ex-covering whose elements are in A_ 1 U i.}. 0 •

46

This is highly suggestive that there may be numerous extensions of

Konig's theorem, and indeed there are.

8.4. If G contains no subgraph in L}.k U A.k+i U ... then G has an ex-covering

whose elements are in A_ 1 U i.}..0 U ... U L}.k-l·

Proof: This is a direct consequence of 8.1. D

At this point one might be tempted to say: in light of 3.2, which

guarantees that the graphs in A2, L}.3, .•. all contain odd cycles, Konig's theorem,

as stated in 8.3, is simply a special case of 8.4 (namely the case k = 1).

But this would be begging the question, for Berge's proof of 3.2 in fact

uses Konig's theorem.

However, Berge's result (3.2) generalizes a couple of earlier results, in

particular one of Beineke, Harary, and Plummer ([BHP], p. 208) which state:

Any pair of adjacent critical edges of a graph

lies on an odd cycle.

Page 53: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

47

By a critical edge, Beineke, Harary, and Plummer mean an edge whose

removal increases a graph's independence number. Their proof of this result

does not rely on Konig's theorem, and when applied to an edge-critical graph

(where all the edges are critical) it may be stated:

Any adjacent pair of edges in an edge-critical graph

lies on an odd-cycle.

This is the same as Berge's result, but without the work "chordless," and

suffices to allow the offhand deduction of Konig's theorem (as above) without

committing the fallacy of petitio principii.

8.4, however, is a terribly weak theorem, for it essentially has infinitely

many hypotheses. Much stronger would be the following.

8.5 (Conjecture) If G contains no subgraph in .O.k then G has an ex-covering

whose elements are in .0._1 U .0.0 U ... U .O.k-1·

As is shown below, 8.5 would follow quickly from the following.

8.6 (Conjecture) If G E .D.k then G has a subgraph H E .O.k-l·

To see that 8.6 implies 8.5, suppose that 8.6 holds, and that the graph G

contains no subgraph in .O.k. So as not to contradict 8.6, G contains no

subgraph in Ak+i- By induction then, G contains no subgraph in Ak+Z• or in

.0.1'+3 , and so on ad infinitum.

Page 54: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

48

Thus G contains no subgraph in l:ik U l:ik+i U ... , and by virtue of 8.4, G

must have an ex-covering whose elements are in /:i_ 1 U 6. 0 U ... U /:ik-l• which is

the conclusion of 8.5.

Such a general result as 8.6 seems at this time to be out of reach. The

next chapter is devoted to the first unresolved case in 8.5, the case k = 2.

Page 55: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

49

CHAPTER 9 - Various Results.

It will be convenient to speak of coverings as including vertices and edges

of a given graph. While vertices and edges are not, strictly speaking, subsets

of a graph, this abuse of the language will not give rise to any ambiguity.

When we speak of a vertex v in this sense, let it be understood that we mean

the subgraph ({v}, ¢), and for an edge e = {u, v}, we mean the subgraph ({u, v},

{e}).

We seek a result of the following kind, for such would indeed be a

genuine extension of Konig's Theorem and a strong result:

9.1 (Conjecture) If G has no subgraph which is a very-odd-K 4 (i.e., element

of A. 2 ), then G has an a.-covering whose elements are vertices, edges, and odd

cycles (ie: elements of A.- 1 U A.0 U A. 1).

To this end we focus our attention on the following slightly more

technical conjecture from which 9.1 would follow immediately:

9.2 (Conjecture) If G E A.10 where k > 2, then G has a subgraph in A.2 •

Unfortunately, our efforts toward this end will not be entirely successful;

that is, 9 .2 will not be proved. These efforts are included, however, for a

number of reasons: they are numerous and exemplify interesting techniques;

they proceed a long way in a direction which appears to be promising; they will

allow us to prove another extension of Konig's Theorem, one which is weaker

than 9.1, and one which is already known, but whose proof relies heavily on

Page 56: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

50

advanced theoretical notions and techniques (most notably the theory of

regular matroids). Presently, we discuss this just-mentioned extension of

Konig's Theorem, while making apparent our choice of the name very-odd-K 4 for

an element of ~2 •

In [G2], Gerards defines an odd-K 4 to be a graph of the type depicted

below:

Illustration 18:

where the wriggled lines denote pairwise vertex disjoint paths, and where the

four "faces" (triangles) of this "tetrahedron" are bound by cycles of odd length.

The graphs which we have called very-odd-K4 's comprise a proper subset

of the collection of odd-K4 's. This may be seen by sorting the odd-K/s

according to the various allowable collections of parities of the numbers of

edges in the six paths denoted by the wriggled lines. Illustration 19 depicts the

three types. The labels o (for odd) and e (for even) refer to the number of

edges in a given path. The odd-K 4 's of type I are precisely the very-odd-K 4 's,

the graphs in ~2 •

Page 57: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

(

51

Illustration 19: The three types of odd-K 4 's.

Gerards continues in [G2] to prove the following extension of Konig's

Theorem, stated here in the terms to which we have become accustomed:

9.3 If G has no subgraph which is an odd-K4 then G has an ex-covering whose

elements are vertices, edges, and odd cycles.

In the next chapter, we will give a graph theoretic proof of 9.3. A subset

of that proof will provide [and this will be duly noted at the proper time] a

proof of the following more technical result.

9.4 If G E l:i.k for k > 2, then G contains a subgraph which is an odd-K4 •

[Compare this to 9.2 which conjectures: If G E Ak for k > 2, then G

contains a subgraph which is a very-odd-K 4 .]

The remainder of this chapter proceeds as follows. We will assume the

existence of a certain graph X which provides a counterexample to the

conjectured 9.2. We will also assume the existence of a certain graph Y which

provides a counterexample to 9.4. We will then let the graph Z be either of

Page 58: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

52

the graphs X or Y, and show that in either case, Z (from whose supposed

existence we wish to derive a contradiction) must have some very particular

properties.

In the next chapter we will see that, if Z = Y, these properties are

sufficient to allow us to arrive at the desired contradiction, thus proving 9.4,

and, as mentioned, Gerards' extension of Konig's theorem (9.3) will follow from

this.

The following result tells us that if there exists a counterexample to 9.2

[respectively, 9.4], then there exists one which is 3-connected, Its proof relies

entirely on Gallai's characterization of the (2 but not 3)-connected edge-critical

graphs given in 3.11.

9.5 If for some k ~ 3 there exists a graph in f::.k which is (2 but not 3)­

connected, and which contains no very-odd-K4 [resp. no odd-K 4], then for some j

with k 2 j 2 3, there exists a graph in f::.J which is 3-connected and contains no

very-odd-K4 [resp. no odd-K4 ].

!:roof: (Refer to 3.11) Suppose that there are edge-critical graphs with o 2 3

which are (2 but not 3)-connected and which contain no very-odd-K 4 's [resp. no

odd-K/s]. Restrict attention to such graphs of minimum o, and from among

these, let G be one with a minimum number of edges, and let o(G) = k

(i.e., G E Ak).

Let G! and G~ be as in 3.ll(b). Refer to Illustration 20, below.

Page 59: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

Illustration 20:

I

I

\

\

I / G_,_

/

odd u-v p<:..-t-h I~ G;

\

\

.1

/

53

Suppose also, and without loss of generality, that of Gj and G2 , G~ is the

one with the property that the graph G1 obtained from it by adding the edge

{u, v} is also in A, and that G2 is the one with the property that the graph G2

obtained from it by identifying u and v is again in A.

Since k = o(G) = 6'(G 1) + o(G 2 ) - 1 (3.1 l(c)), the possibilities for the pair

(o(G 1), o(G)) are (k, 1), (k - 1, 2), (k - 2, 3), ... , (2, k - 1), (1, k).

(i) G 1 contains no very-odd-K 4 [resp. odd-K,i} for: if G1 contains such a

subgraph not using the edge {u, v}, then this subgraph would also be a

subgraph of G; and, if G1 contains such a subgraph using the edge {u, v},

then, in G, this subgraph, with {u, v} replaced by a path of odd length

from u to v lying entirely in G2 , would contradict G containing no very-

odd-K4 [odd-K4]. To see that such a path exists, let e 1 and e 2 ,

Page 60: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

respectively, be incident with u and v in G2 . In G2, where u and v are

identified, they are adjacent, and by virtue of 3.2, they lie on a

chordless odd cycle. This odd cycle in 0 2 is the desired path of odd

length in G2 .

(ii) G2 contains no very-odd-K 4 [odd-K4 ], for: if G2 contains such a subgraph

not using the vertex which results from the identification of u and v,

then this subgraph would also be a subgraph of G; and if 0 2 contains

54

such a subgraph which does use this vertex, then, in G, this subgraph,

with u and v split, together with a path of even length in Gi connecting u

and v would again contradict G's containing no very-odd-K 4's [odd-K.1's].

The existence of such a path is guaranteed, again by virtue of 3.2, for if

e3 is an edge of G1 not equal to {u, v} but incident with u, then in 0 11 the

adjacent edges e3 and {u, v} must lie on an odd cycle. The removal of the

edge {u, v} from this cycle provides the desired path of even length from

u to v in Gf .

The minimality properties of G imply that if o(G;) > 2 then G1 will be 3-

connected, and will therefore provide the graph whose existence is asserted in

the conclusion of the theorem. If neither 6(01) > 2 nor 6(0 2) > 2, then it must

be the case that o(G1) = 6(G 2 ) = 2 in which case G1 E ~2 and G2 E ~2, and the

arguments (i) and (ii) above, ensure that G (in ~3) has a subgraph in ~2, which

is contrary to hypothesis. D

In light of 9.5, in trying to show that every connected edge-critical graph

with 6 > 2 must contain a very-odd-K 4 [odd-K4] we may restrict our attention to

those graphs which are 3-connected. That is, if there exists a counter-example

to 9.2, [or to 9.4], then there exists one which is 3-connected.

Page 61: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

55

We now suppose the existence of counterexamples to 9.2 and 9.4, fixing

the graphs X and Y for the remainder of the chapter.

If there exists a counterexample to 9.2, let it have the following

properties (without loss of generality), and let it be called X:

(a) X is edge cri ti ca 1

(b) X is 3-connected

(c) 6(X) > 2

(d) X contains no very-odd-K 4

(e) 6(X) is minimum among all graphs with properties (a), (b), (c), and (d)

(f) e(X) is minimum among all graphs with properties (a), (b), (c), (d),

and (e).

If there exists a counterexample to 9.4 let it have the following

properties (without loss of generality), and let it be called Y:

(a) Y is edge-critical

(b) Y is 3-connected

(c) 6(Y) > 2

(d) Y contains no odd-K 4

(e) 6(Y) is minimum among all graphs with properties (a), (b), (c), and (d)

(f) e(Y) is minimum among all graphs with properties (a), (b), (c), (d),

and (e).

Let Z be either of the graphs X or Y. The propositions which follow are

true if Z = X or if Z = Y. The proofs are written as if Z = X. Where a

different argument is needed in the case Z = Y, it is given in [bracketed]

remarks.

Page 62: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

56

9.6 For every vertex v of Z, p(v) ~ 3.

Proof: If some vertex v of Z has degree 2, then N(v) would comprise a cut-set

of Z of cardinality '.:, which is impossible since Z is 3-connected. 0

In [S3J, Suranyi has proved the following lemma, which we will exploit

greatly, and which is stated below in terms appropriate to this discussion.

9.7 (Suranyi, [S3J, p. 1416) If Z E Ilk for k ::e: 1, and if x is a vertex of Z,

then there exists an edge-critical subgraph H, spanning Z - x, containing no

isolated vertices and satisfying oc(H) = oc(Z) ( = a.(Z - x)).

The subgraph H is not unique. In our discussion we will let any

realization of this subgraph be denoted Z - x.

Observe that

o(Z - x) = v(Z - x) - 2a.(Z - x) = v(Z) - 1 - 2a.(Z) = o(Z) - 1.

9.8 The components of Z - x are edges and odd cycles.

Proof: Since Z - x has no isolated vertices, its components lie in

tl 0 U fl 1 U ... U fl 0czi-i· These components contain no very-odd-K4's, [odd-K/s],

for if one did, so also would Z, and thus by the minimality conditions, (e) and

(f) on Z must lie in !l0 U fl 1• D

o(Z - x) is equal to the number of odd-cycles in Z - x because of the

Page 63: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

57

additive nature of 8 on the components of a graph, and so Z - x is a collection

of precisely o(Z) - 1 disjoint odd cycles, and perhaps some isolated edges.

As mentioned earlier, we will proceed to show that, for a given vertex x

of Z, and a given realization of the graph Z - x, the graph Z - x must satisfy

certain properties. Many of the arguments in the remainder of this chapter

take the following form: A certa.in covering of Z - x by precisely o(Z) - 1

disjoint odd cycles and the appropriate number of isolated edges cannot exist,

for if, as it did, it would provide a realization of possessing a

property which we have shown, it must possess.

9.9 The neighbors of x in Z lie on the odd cycles of Z - x.

Proof: Suppose that, on the contrary, {x, y} is an edge of Z and that {y, w} is

one of the isolated edges of Z - x. In Z, y has yet another neighbor, say z,

since p(y) :?: 3.

Illustration 21:

- -.... / '-

/ " I _j w \

I \

' ' r-/1/-' ....

' \ '9 .z:

/ \ I

' ' / ....... /

/

l.

Page 64: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

58

a.CZ - {y, z}) > a.(Z) since Z is edge-critical. But an o:-set I of Z - {y, z}

is necessarily independent in Z - x for containing both y and z, it cannot

contain x. Thus I is independent in Z - x and of cardinality

o:(Z - {y, z}) > a.(Z) = o:(Z-x), which is absurd. D

9.10 If y is a neighbor of x then p(y) = 3, y having no neighbors other than x

and its two immediate neighbors on that odd-cycle.

Proof: Suppose, as is depicted in Illustration 22, that some neighbor y of x has

as a neighbor w, not an immediate neighbor on the cycle of Z - x to which y

belongs. Two cases are pictured. No distinction is needed in the proof.

Illustration 22:

.w -(w)

-- -

Just as in the last paragraph of the proof of 9.9, an ex-set of Z - {y, w}

would be independent in Z - x and of cardinality greater than cx(Z - x), which is

impossible. 0

9.11 For every vertex v of Z, p(v) 3.

Page 65: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

59

Proof: Let v be a vertex of Z and let x be a neighbor of v. By 9.9 and 9.10,

in any subgraph of Z - x which qualifies for being called Z - x, v lies on an

odd-cycle of Z - x and has no neighbors other than x and its two immediate

neighbors on this cycle. 0

Proposition 9.11 implies that Z is 3-regular (!), a very strong property.

9.12 o(Z) is even and consequently greater than or equal to 4.

Proof: The sum of the degrees of the vertices of a graph must be even (being

twice the number of edges - a well known elementary result). Since in Z all

the degrees are 3, 3v(Z) must be even, requiring that v(Z) be even and

consequently that o(Z) = v(Z) - 2a.(Z) be even. D

9.13 If x has two neighbors y and z on the same odd cycle of Z - x, then on

this cycle, y and z have a common neighbor.

Proof: Suppose the contrary, with x, y, and z as above.

Page 66: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

Illustration 23:

Ii ew 1°':;)0/o.. +ed .e~e.s

60

Then y and z divide this cycle into two y - z paths, one containing an

odd number of edges, the other containing an even number, greater than 2, of

edges. Let w be adjacent to z and on the second of these paths. The

remainder of the vertices along this path may be paired, as is shown by the

wriggled lines.

Now consider the odd cycle formed by the first of the above mentioned

paths and the edges {x, y} and {x, z}, together with the pairs of vertices just

mentioned. They together include the same number of vertices as did the

original cycle of Z - x (for x is now included, while w is omitted), and have the

same value of v - 2o:. as did that cycle, namely 1. Consequently the subgraph

consisting of this cycle and these edges has the same 6 as did the original cycle

of Z - x.

Now if, in the collection of edges and odd cycles which comprises Z - x,

we replace the original odd cycle with this new one and these isolated edges, we

Page 67: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

61

get a graph qualified to serve as Z - w but with w having a neighbor not on

one of the odd cycles of Z - w, contrary to 9.9. 0

In particular, 9.13 implies the following two results.

9.14 Z contains no triangles.

Proof: Suppose Z contains a triangle whose vertices are x, y, and z (see

Illustration 24). y and z must lie on cycles of Z - x and since each of these

cycles must contain the edge {y, z}, they must be the same cycle.

Illustration 24:

( '

odd path

\.._ ____ ~, '- ........ __ ......

So as not to contradict 9.13, y and z must share a neighbor w on this

cycle, requiring it to be a triangle as well. 3.2 demands that in Z, the edges

{x, z} and {z, w} lie on a chordless odd cycle. This cycle, being chordless, must

necessarily a void the vertex y. This implies the existence of a pa th of an odd

number of edges from x to w, not including the vertex y. But the subgraph of

Z now depicted in Illustration 24 is indeed a very-odd-K4 [odd K4]; it is K4 with

precisely one of its edges replaced by a path of odd length. This is contrary

to our choice of Z. D

( 9.15 For no vertex x of Z and no realization of Z - x does x have all three

Page 68: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

62

of its neighbors on the same odd cycle of

Proof: It is impossible for each of the 3 pairs of neighbors of x to have a

common neighbor on the same odd cycle unless that cycle is a triangle

containing only them, and this by 9.14 is impossible. D

9.16 For any vertex x of Z there exists a realization of - x for which no

edge of Z is a chord of the cycles of - x which neighbor x.

Proof: Let x be any vertex of Z, and choose any Z -x for which the sum of

the lengths of the cycles neighboring x in Z is a minimum. If there exists an

edge e of Z that is a chord of a cycle C of Z -x neighboring x in Z, then the

endpoints of e divide C into two arcs, one with an even number of edges, and

one with an odd number of edges.

Illustration 25:

-t-h~ jde.. c This even arc together with e forms an odd cycle, while the vertices

along the arc with an odd number of edges may be divided into adjacent pairs.

In Z - x, the cycle C can be replaced by this new odd cycle and the edges

formed by these adjacent pairs of edges.

Page 69: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

(i) If x has a neighbor on the arc with an odd number of edges, then in

this new version of Z - x, x neighbors an isolated edge, which is

contrary to 9.9.

(ii) If x has a neighbor on the even arc, this new version of Z - x

violates the minimality condition above. 0

63

9.17 For at least one choice of the vertex x of Z, and for at least one

realization of the subgraph Z, x has as neighbors a vertex on each of

three distinct odd cycles of Z - x.

Proof: Suppose on the contrary that for every choice of the vertex x, and

every realization of Z - x, it happens that x has as neighbors two vertices on

one of the odd cycles of Z - x. Notice that 9.13 implies that these two

neighbors of x have another common neighbor on that cycle, and that 9.15

implies that the third neighbor of x be on some other odd cycle of Z - x.

The proof proceeds by induction, giving a construction which, under our

suppositions, continues ad infinitum, implying Z has an infinite number of

vertices, which is not the case.

Reference to Illustration 26 will be indispensable to the arguments which

follow. In the other illustrations in this proof, let the notation ** following

the illustration number indicate that all else, not pictured, but whose existence

has at that point been asserted, is as it is in Illustration 26.

Page 70: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

Illustration 26:

\

\

\

\

64

c..:i..

Ci;.

Page 71: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

65

Let x be a vertex of Z, and from among all the realizations of

which the odd cycles neighboring x are chordless in Z (see 9.16), choose one for

which the number of isolated edges is a minimum. Suppose that its connected

components are the odd cycles Ci, i = 1, 2, ... , 8(Z) - 1, and that the isolated

By hypothesis, x is adjacent to vertices u 2 and v 2 on one of the odd

cycles, call it C2 , and to u 1 on C 1• By 9.13, u 2 and v 2 have a common neighbor

w2 on C2 • Since p(w 2 ) = 3, w2 has yet another neighbor, w 3 which we will

presently show must lie on an as yet unmentioned cycle of Z - x which we will

Observe that w 3 is not on the cycle C 1, for this would imply that Z contains a

very-odd-K 4 [odd-K4 ], namely the one shown in Illustration 27, where the dashed

line denotes the arc along C1 from w3 to u 1 that has an odd number of edges.

Illustration 27: (Very-odd-K4 if w3 is on C1)

c~

Page 72: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

66

Moreover, w3 is not on C2 , for we have taken C2 to be chordless.

Lastly, w3 is not on one of the isolated edges of Z - x, for if it were,

there would exist a realization of Z-::-w2 , for which, in Z, w2 would neighbor

one of the isolated edges (see Illustration 28). This is contrary to 9.9.

Illustration 28**: (w3 on an isolated edge of Z - x: realization of Z - w2

contradicting 9.9)

Thus w3 must lie on a previously unmentioned cycle of Z - x which we

will call C3 •

At this point we have just completed the initial step of the induction

argument. We will proceed to show that given a vertex Wzn+1' with certain

properties which will be stated presently, it must be the case that Z has

numerous previously unmentioned additional vertices, and in particular one

among them which we will call Wz 71 +3, having the very same properties which

( allowed the deduction of its existence from the existence of Wzn+i· This will

imply that Z has an infinite number of vertiees, w3 , w5 , .. ., which is absurd, and

Page 73: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

67

when we reach this implication, the proof will be complete.

(The induction hypothesis) Let us suppose that we have shown the

existence of w2 n+ 11 being on a previously unmentioned odd cycle C2 n+i of

and being a neighbor of the previously mentioned vertex w2 ,,, lying on C2 ,,,

where W2n has neighbors U2n and V2n On C2n• and Uz 71 and V2n also have the

common neighbor x 2 n.

Observe that if we let x also be called x2 , then w3 satisfies the induction

hypothesis in the case n = 1.

For the sake of clarity, we present the induction step in terms of the

subscripts of the very next iteration, that is, the deduction of the existence of

w5 , satisfying the induction hypothesis, from the existence of w3 • The very

same arguments, with the appropriate subscripts, will allow the deduction of the

existence of w2,,+3 from that of w2n+i·

Continuing, w3 has two neighbors u 3 and v 3 on C3 • By hypothesis, in any

realization of Z - w 3 , two of the neighbors in Z of w3 have yet another common

neighbor. w2 cannot be one of these two, for neither of w2's neighbors (u2 and

v 2 ) neighbor u 3 or v 3 • So u 3 and v 3 have a previously unmentioned vertex x3

as a neighbor.

Now x3 must lie in some component of Z - x. Since each vertex has in Z

degree 3, x3 can not lie on C1 or C2 • Neither can x3 lie on C3 , for if it did, it

would have to be adjacent to u 3 (or v 3 ) along C3 , giving rise to a realization of

Z - w 2 in which in Z, w 2 neighbors one of the isolated edges of

Page 74: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

68

Z - w2 , which cannot happen. This realization of Z - w2 is depicted in

Illustration 29.

Illustration 29**: (x3 on C3 : realization of Z - w2 contradicting 9.9)

C..z. u.%.

Vi..

\ \

\

1C::. /)(._z \ . 'X-3 U.3

WJ c.3

Also, this component of Z - x in which x3 lies cannot be yet another cycle

of Z - x, for two of the neighbors of x 3 already lie on C3 • Thus x 3 must lie on

one of the isolated edges, say {x3 , x 4}, of Z - x. ·

Consider the realization of Z - x4 depicted in Illustration 30.

Page 75: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

69

Illustration 30**: (A realization of Z - x 4 )

In this version of Z - x 4 , two of the neighbors of x4 must share yet

another common neighbor. If one of these two is x3 , then the other must be y 3

or z 3 • But z 3 (and similarly y 3 ) neighboring x4 would provide a realization of

Z - x contradicting the condition that our original Z - x has a minimum number

of isolated edges. This realization of Z - x is shown in Illustration 31.

(

Page 76: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

70

Illustration 31 **: (x 4 neighboring z 3 (or y 3): contradictory realization of Z - x)

c.3

Note that the edge {v 3, z 3} being in Z a chord of the pictured "new" C3 ,

causes no difficulty, for this is not one of the cycles neighboring x.

So x 4 has two "new" vertices as neighbors, call them u 4 and v 4 , and these

two have yet another common neighbor, call it w4 •

If we consider the realization of Z - x 4 depicted in Illustration 30, we see

that the vertices u 4 , w'l, and v 4 must lie on some previously unmentioned cycle

of Z - x, call it C4 •

Furthermore, w4 must have yet another neighbor, call it w5 • Now, w5 is

not on C1, C2 , or C3, since this would imply the existence of a subgraph of Z

which is a very-odd-K 4 [odd-K 4 ], (just as w3 could not be on C1), as shown in

Illustration 32.

Page 77: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

71

Illustration 32: (Very-odd-K 4 if w5 is on C 1, C2, or C3 )

Nor can w5 be on C4, for this would imply the existence of a realization

of Z - x .. 11 shown in Illustration 33, for which in Z, x4 neighbors one of the

isolated edges (compare with the proof of 9.13 and Illustration 23).

Page 78: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

Illustration 33**: (w 5 on C 4 : realization of Z - x 4 contradicting 9.9)

c'l.

\

w~/ Finally, w5 cannot be on one of the isolated edges of Z - x (as w3 could

not have been), for there would then exist a realization of Z - w4 , as depicted

in Ill ustra ti on 34 in which in Z, w 4 neighbors one of the isolated edges,

contrary to 9.9.

(

Page 79: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

73

Illustration 34**: (w5 on an isolated edge of Z - x: realization of Z - W 4

contradicting 9.9)

C::i..

....

Thus w5 lies on a previously unmentioned· odd cycle of Z - x, call it C5 •

The vertex w5 satisfies the induction hypothesis, as did w3 • This implies

the existence of w 7 , w8 , ... , which is absurd. The proof is complete. 0

Page 80: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

74

By virtue of the preceding remarks, we are finally at liberty to choose a

vertex x of Z, and a realization of Z - x, as depicted in Illustration 35, with

the property that in Z, x neighbors three distinct odd cycles of Z - x.

Illustration 35: (with possibly isolated edges and additional odd cycles).

CL_ ·~ \

(

Page 81: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

75

CHAPTER 10 - Gerards' Extension of Konig's Theorem.

We are now in a position to give an elementary proof of Gerards'

extension of Konig's Theorem stated in 9.3 and repeated below:

(9.3) If G has no subgraph which is an odd-K,, then G has an O'_-covering

whose elements are vertices, edges, and odd cycles.

Proofs [of 9.3 and 9.4]: Suppose, on the contrary, there exist graphs which

contain no odd-K 4 but which do not have a-coverings by vertices, edges, and

odd cycles. Let G be one with a minimum number of edges.

As noted earlier, if {Hi> H2, •.• , Hk} is any collection of subgraphs of G

k with V(G) = U V(H;) then

!=l

k

a(G) s 2)x.CHi) i=l

since each a-set of G induces an independent set in each Hi·

If, for some edge e of G, a(G - e) = a_(G), then the graph G - e would

contain no odd-K 4 , would have no ex-covering by vertices, edges, and odd cycles

(since such would be an a-covering of G), and would have fewer edges than G.

This is contrary to our choice of G, and thus, for every edge e of G,

a.(G - e) > ex.CG). That is: G is edge-critical.

The graphs in A_ 1, A0 , and A 1 clearly have a-coverings of the desired

kind, and the graphs in A2 being very-odd-K 4's contain odd-K 4's. So 6(G) > 2.

Page 82: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

76

It follows that there exists a connected edge-critical graph with 6 > 2, and

containing no odd-K 4 • [The contradiction which arises from the existence of

such a graph will not only complete this proof but also provide a proof of 9.4,

mentioned earlier, and again later in this chapter.]

In 9.5 it was shown that if such a graph exists, then one exists which is

3-connected.

It now follows that there exists a graph satisfying the conditions defining

the graph Y of Chapter 9, and in Chapter 9 (see 9.17) it was shown that such a

graph necessarily contains three disjoint odd cycles, CH C2 , and C3 , and a

vertex x adjacent to each of these three cycles, as shown.

( Illustration 36:

C.;. C3

) ~/

One version of Menger's Theorem ([MJ, see [BJ), a well known result in

graph theory, states:

Page 83: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

77

It is a necessary and sufficient condition for a graph to

be k-connected that between any two vertices there exist k

disjoint paths.

It is claimed that (possible after a renaming of the cycles) there exist

such paths as are represented by the dashed lines from z 1 to z 2 and from z2 to

z 3 in lllustration 37.

Menger's theorem guarantees the existence of three x - y 2 paths. Clearly

one of these paths is the edge {x, y 2}. We may assume without loss of

generality that another of these paths avoids one of the cycles C1 and C3 , say

C3 • If the third of them avoids the cycle C1 then the claim is verified; if it

does not, then it visits C 1 before (proceeding from y 2 to x) it visits C3 • In this

case, a renaming of the cycles where C1 is renamed C2 verifies the claim.

Illustration 37:

"- LC ,,, 3 I J '2. " / ' / ' / " / '\. I /

'-.,. I /

'f'-x.

The vertices where these paths meet C2, together with x's neighbor on C2,

divide C2 into three arcs. There are four cases to be considered, according to

Page 84: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

78

the parities of the numbers of edges in each arc.

Case 1: The three arcs are odd. Here, we immediately find an odd-K 4 of type

I, (a very-odd-K 4 ), shown below, which is a contradiction.

Illustration 38:

- ,, - - - ...._

I

'

~3 ...... ,

\

I' /

\ \

I

In Illustration 38, the paths from 2 1 to y 1 and from 2 3 to y 3 are arcs along

C1 and C3 , respectively, chosen so that the 2 2 - 2 1 - y 1 - x and 22 - z 3 - y 3 - x

paths each have an odd number of edges.

Case 2: The arc from z 2 to 22 is odd, the other two even.

Page 85: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

Illustration 39: ~,

I

\

~ I

...,,. - - - .........

odd

-r.. 3 -.,

/

\

\ I

I

79

In this case, we get an odd-K4 of type II (shown below) by choosing the

arcs along C1 and C3 so as to make the z 2 - z 1 - y 1 - x and z~ - z 3 - y 3 - x

paths even. This again is a contradiction.

Illustration 40:

Cases 3 and 4: The arc from y 2 to z 2 (and by a symmetrical argument, the arc

from y 2 to z9 is odd, the other two even.

Page 86: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

80

Illustration 41:

~I ever; 3 ~ ,,,.- - - - - -.,

I ~ z;\ \

I \

odd evej l

I \ ,, ' /

~) _j3

In this case, the arcs along C1 and C3 are chosen so as to make the

z 2 - z 1 - y 1 - x path odd and the z2 - z 3 - y 3 - x path even, yielding the odd-K4

of type III shown below, and also providing the contradiction which completes

the proof.

Illustration 42:

D

To close this chapter, we recount what has indeed been shown in our

attempt to prove conjecture 9.2 (and consequently, conjec:ture 9.1).

Page 87: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

81

(9.4) If G is connected and edge-critical with o(G) > 2, then G contains as a

subgraph an odd-K 4 •

Proof: See the proof of 9.3 given earlier in this chapter. 0

10.1 If G is connected and edge-critical with o(G) = 3, then G contains as a

subgraph a very-odd-K4 (an element of Ll 2). [This is the first unsettled

instance of conjectures 9.2 and 8.6.]

Proof: It is a direct consequence of 9.12, which, together with the development

preceding it implies that if G is connected and edge-critical, not containing a

very-odd-K 4, and of minimum o among such graphs with o > 2, then o(G) ~ 4 (and

even). D

10.2 If G is connected and contains no very-odd-K4 and v(G) - 2cx.(G) ~ 3, then

G has an cx.-covering by vertices, edges, and odd cycles.

Proof: Suppose not, and let G be a counterexample with a minimum number of

edges. Then (as in the proof of 9.3) G is edge-critical. If o(G) < 2, G is an

isolated vertex, an isolated edge, or an odd cycle, a contradiction in any case.

If o(G) = 2, G is a very-odd-K4 , again a contradiction, while if o(G) = 3, G

contains a very-odd-K 4 by virtue of 10.2 which is once again a contradiction. 0

Page 88: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

CH 1.PTER 11 - A note on a theorem of Graver and Yackel

A tree is a connected graph containing no cycles.

Graver ([G3J) has suggested the following notions:

Let G = (V, E) be a tree. Define a function hG with domain V U E and

codomain {O, 1} as follows:

a) for v E V,

b) for e = {u, v} E E,

if v E I for every ex-set I of G

otherwise;

if {u, v} n I rf- 0 for every cx.-set I of G, while hG(u)

otherwise.

0

82

It is this very notion, that of the function hG, which has proved to be

the seed from which grew all of the work in this thesis.

Note also that, since a tree has no cycles, a preferred ex-decomposition

(Chapter 5) of a tree contains only vertices and edges (elements of LL 1 U A0 ).

A moment's reflection reveals that the preferred ex-decomposition of G is

in fact unique, and consists of precisely those edges and vertices for which the

value of hG is 1, i.e., the support of hG.

Page 89: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

83

The function hG may also be defined as follows:

(i) for v E V, hG(v) = cx(G) - cx(G - v), and

(ij) for e = {u, v} E E,

That the two definitions of hG agree on the vertices of G is clear. To see that

they also coincide on the edges of G, observe that hG(e), as originally defined

(1) equalling 1 implies cx(G) - cx(G - u - v) = 1 while hG(u) = hG(v) = O;

(2) equalling 0 implies either

(a) some a.-set of G misses both u and v, whence

cx(G) - a.CG - u - v) = 0 and hG(u) = hG(v) = 0, or

(b) every o:.-set of G includes u (similarly, v), in which case

cx(G) - cx(G - u - v) = 1 while hG(u) = 1 and hc(v) = 0.

In [GY] (p. 137), Graver and Yackel have shown that for a tree G = (V, E),

where gi is the number of connected subgraphs of G containing i edges.

Our second way of defining the function hG allows us to prove the

following corollary to this theorem of Graver and Yackel, and with this

proposition and its proof, we end this work.

(

Page 90: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

84

11.1 Let G = (V, E) be a tree. For a vertex v of G (respectively, an edge e

of G) let g 1(v) (respectively, g1(e)) denote the number of connected subgraphs of

G c0ntaining the vertex v (respectively, the edge e), and having i edges. Then

IEI . hG(v) = 2:::C-U'g,(v)

i=O

and

i=l

Proof: Let v be a vertex of G. By the above remarks

hG(v) = a_(G) - cx(G - v)

where E' = E(G - v) and g~(v) is the number of connected subgraphs of G - v

containing i edges.

A connected subgraph of G not containing the vertex v contributes 0 to

the above difference, since it contributes equally to each of the two

summations. On the other hand, connected subgraphs of G co1itaining v are not

even considered in the second summation.

Thus the above difference is equal to

IEI . 2.::(-1/g;(v) as desired. i=O

Page 91: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

85

Furthermore,

where E' = E(G - u - v), and g~ is the number of connected subgraphs of

G - u - v containing i edges.

Now a connected subgraph of G containing i edges but neither u nor v

contributes zero to the above, being counted equally in the first two

summations, and not at all in the last two.

Also, a connected subgraph of G containing i edges and u but not v

(similarly v but not u) contributes zero, being counted equally in the first and

third summations (or first and fourth) and not at all in the other two.

The only remaining connected subgraphs of G are those containing i edges

and both u and v. But such must contain the edge e for, G having no cycles,

the only path from u to v is along that edge. Also, such a subgraph

contributes equally to the first, third, and fourth summations, and not at all to

the second. Consequently, the above equals

!El . - I:C-lYgi(e),

i=l

Page 92: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

86

the summation now beginning at i = 1 since there are no connected subgraphs of

G containing u and v and no edges. This is as desired. 0

Page 93: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(

BIBLIOGRAPHY

[A] B. Andrasfai: "On line-critical graphs", Theory of Graphs International

Symposium, Rome, 1966; Dunod, Paris-Gordon, and Breach, New York,

1967, 9-19.

[B] C. Berge: "Graphs and Hypergraphs'', North-Holland, Amsterdam­

London, American Elsevier, New York, 1970.

[BHP] L.W. Bein eke, F. Harary, and M. Plummer: "On the critical lines of a

graph", Pacific Journal of Mathematics, 22, no. 2, (1967), 205-212.

[BM] J .A. Bondy and U .S.R. Murty: "Graph Theory with Applications",

American Elsevier, New York, 1976.

[Gl] A. George: "On line-critical graphs", (Master's thesis), Vanderbilt

University, August 1971.

[G2J A.M.H. Gerards: "An extension of Konig's theorem to graphs with no

odd-K/', (preprint).

(G3] J.E. Graver: personal communication.

[GW] J.E. Graver and M.E. Watkins: "Combinatorics with emphasis on the

theory of graphs", Springer-Verlag, New York, 1977.

87

Page 94: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

88

[GY] J.E. Graver and J. Yackel: "Some graph theoretic results associated

with Ramsey's theorem", Journal of Combinatorial Theory, vol. 4, no. 2,

(1968), 125-175.

[Hl] A. Hajnal: "A theorem on k-saturated graphs", Canadian Journal of

Mathematics, 7, (1965), 720 - 724.

[H2] F. Harary: "Graph Theory'', Addison-Wesley, Reading, Mass., 1969.

[Kl D. Konig: "Theorie der endlichen und unendlichen Graphen", Chelsea,

New York, 1950.

[Ll] L. Lovasz: "Some finite basis theorems in graph theory", Colloquia

Mathematica Societatis Janos Bolyai 18, Combinatorics, Keszthely,

Hungary, (1976), 717 - 729.

[L2] L. Lovasz: "Combinatorial Problems and Exercises'', North-Holland,

New York, 1979.

[M] K. Menger: "Zur allgemeinen Kurventheorie", Fund, Math., 10, (1926),

96-115.

[Sl] L. Suranyi: "A note concerning a conjecture of Gallai concerning

a-critical graphs", Colloquia Mathematica Societatis Janos Bolyai 18,

Combinatorics, Keszthely, Hungary, (1976), 1065-1074.

Page 95: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

[S2J L. Suranyi: "On a generalization of line-critical graphs", Discrete

Mathematics 30, (1980), 277-287.

[S3J L. Suranyi: "On line-critical graphs", Colloquia Mathematica Societatis

Janos Bolyai 10, Infinite and Finite Sets, Keszthely, Hungary, (1973),

1411-1444.

[W] M.E. Watkins: "On the existence of certain disjoint arcs in graphs",

Duke Mathematical Journal, vol. 35, no. 2, (1968), 231-246.

[ZJ A.A. Zykov: "On some properties of linear complexes", (Russian)

Mathematiceskii Sbornik N.S. 24(66), (1949), 163 - 188.

89

Page 96: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

90

INDEX

adjacent ............................. 5 incident ................................. 5 o:.(G) ................................... 9 independence number .............. 9 a-covering ....................... 42 independent ........................... 8 a.-decomposi ti on ................ 20 independent set ...................... 9 ex-set ................................. 9 isolated vertex .................... 11 a.-support ......................... 20

K,, ......................................... 6 k-connected ......................... 14

bipartite graph ................. 44 Konig's Theorem ............. 44, 45

C,, ...................................... 7 maximum independent set ........ 9 chord ................................. 8 circuit ............................... ? complete graph ................... 6 connected ........................... 7 N(v) ...................................... 8 connectivity ..................... 14 neighbor ................................ 5 covering ........................... 42 neighbor set .......................... 8 cube ................................ 26 cut-set ............................. 11 cycle ................................. 7 octahedron, .......................... 25

odd-K 4 ................................. 50 odd subdivision of K 4 ........... 12

deficiency .......................... 9 degree ................................ 7

D.. ..................................... 12 L'.ln ................................... 12 path ...................................... 7 o(G) ................................... 9 platonic graph ...................... 24 dodecahedron .................... 28 preferred ex-decomposition .... 20

E(G) ................................... 5 p(v) ....................................... 7 e(G) ................................... 9 edge ................................... 5 edge-critical graph ............ 10 span .................................... 11 endpoint ............................. 5 subgraph ............................... 8

finite basis theorem .......... 13 tetrahedron ......................... 24 tree ..................................... 82

G - e ................................. 8 G - S ................................. 8 V(G) ...................................... 5 G - v ................................. 8 v(G) ...................................... 9 graph ................................. 5 V a(G) .................................. 20

valence .................................. 7 vertex ................................... 5

hG .................................... 82 very-odd-K 4 ......................... 12 hereditary ........................ 18

Z - x ................................... 56 icosahedron ...................... 26

Page 97: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

BIOGRAPHICAL DAT A

VINCENT EDWARD FA TICA

Date and Place of Birth

21 February 1949

Albany, New York

Elementary School

Saint Patrick's School

Ravena, New York

High School

The Vincentian Institute

Albany, New York

Higher Ed uca ti on

B.S., Syracuse University 1972

M.S., Syracuse University 1977

M.Ph., Syracuse University 1985

91

Page 98: On Edge-Critical Graphs and the Notion of Vertex Independence In Graphs

(


Recommended