Theorem 5.

13 If G is a graph of order n and size m ≥ n − 1, then

If G Theorem 5.14 If G is a connected graph of order at least 3, then its square G2 is 2-connected.
Link: https://math.stackexchange.com/questions/1965914/power-of-a-graph-proof
Theorem 5.15 For every two integers r and n with 2 ≤ r < n,

5.4 Menger’s Theorem

YT Link: https://www.youtube.com/watch?v=2rbbq-Mk-YE
There are many theorems in mathematics which state that the minimum number of elements in some set
equals the maximum number of elements in some other set. The following theorem is such a “minmax”
theorem. It is referred to as Menger’s theorem.
Theorem 5.16 (Menger′s Theorem) Let u and v be nonadjacent vertices in a graph G. The minimum
number of vertices in a u − v separating set equals the maximum number of internally disjoint u − v paths
in G.
Case 1. There exists a minimum u − v separating set U in G containing a vertex x that is adjacent to both
u and v.
Case 2. There exists a minimum u − v separating set W in G containing a vertex in W that is not adjacent
to u and a vertex in W that is not adjacent to v.
Case 3. For each minimum u − v separating set S in G, either every vertex of S is adjacent to u and not
adjacent to v or every vertex of S is adjacent to v and not adjacent to u.

Fun fact:
As we mentioned above, Theorem 5.16 is referred to as Menger’s Theorem, named for Karl Menger who
was born in Vienna, Austria on January 13, 1902. Menger developed a talent for mathematics and physics
at an early age and entered the University of Vienna in 1920 to study physics. The following year he
attended a lecture by Hans Hahn on Neueres über den Kurvenbegriff (What’s new concerning the concept
of a curve) and Menger’s interests turned towards mathematics. In the lecture it was mentioned that there
was no satisfactory definition of a curve (at that time), despite unsuccessful attempts to do so by a number
of distinguished mathematicians, including Georg Cantor, Camille Jordan and Giuseppe Peano. Some
mathematicians including Felix Hausdorff and Ludwig Bieberbach felt that it was unlikely that this
problem would ever be solved. Despite being an undergraduate with limited mathematical background,
Menger solved the problem and presented his solution to Hahn. This led Menger to work on curve and
dimension theory.
Menger became quite ill while a student and his studies were interrupted. During this time both of his
parents died. He eventually returned to the university where he completed his studies under Hahn in 1924.
The next year Menger went to Amsterdam to work with Luitzen Brouwer and Menger broadened his
mathematical interests. In 1927 Menger returned to the University of Vienna to accept the position of
Chair of Geometry. It was during that year that he published the paper “Zur allgemeinen Kurventheorie”
(Menger’s theorem). Menger himself referred to this result as the “n-arc theorem” and proved it as a
lemma for a theorem in curve theory.
In the spring of 1930, Karl Menger traveled to Budapest and met many Hungarian mathematicians,
including Dénes König. Menger had read some of König’s papers. During his visit, Menger learned that
König was working on a book that would contain what was known at that time about graph theory.
Menger was pleased to hear this and mentioned his theorem to König, which had only been published
three years earlier. König was not aware of Menger’s work and, in fact, didn’t believe that the theorem
was true. Indeed, the very evening of their meeting, König set out to construct a counterexample. When
the two met again the next day, König greeted Menger with “A sleepless night!”. König then asked
Menger to describe his proof, which he did. After that, König said that he would add a final section to his
book on the theorem, which he did. This was a major reason why Menger’s theorem became so widely
known among those interested in graph theory.
Because of the political situation in Austria in 1938, Menger left for a position in the United States at the
University of Notre Dame. While there, his work emphasized geometry, which didn’t have the impact of
his earlier work since geometry was not a subject of great interest to many, especially in the United
States. Menger went to the Illinois Institute of Technology in 1948 and spent the rest of his life in the
Chicago area. Menger was considered one of the leading mathematicians of the 20th century. He died on
October 5, 1985. With the aid of Menger’s theorem, Hassler Whitney was able to present a
characterization of k-connected graphs.
We have seen for integers r and n with 2 ≤ r < n that the Harary graphs Hr, n have minimum degree r,
order n and size m with (G) = r = 2m/n . These graphs were defined in terms of powers of cycles. We also
mentioned that the square of every connected graph of order 3 or more is 2-connected. In fact, if k and n
are integers with 2 ≤ k < n and G is a connected graph of order n, then Gk is kconnected (see Exercise
5.39).
In particular, for every connected graph G of order n ≥ 3 and every two distinct vertices u and v of G,
there exist in G2 two internally disjoint u − v paths P and P′. Of course, each edge of P and P′ belongs
either to G or to G2 − E(G). Label each edge of G2 that belongs to G with the label 1 and label each edge
of G2 not belonging to G with the label 2. In general, an edge uv of Gk is labeled i (1 ≤ i ≤ k) if dG(u, v)
= i. Such a graph Gk is then called a distance-labeled graph. A path P in Gk is called proper if every two
adjacent edges in P have different labels. Later we will see that this is related to the topic of graph
colorings (discussed in Chapter 10). While for every connected graph G of diameter 2 or more, the graph
G2 contains a proper u − v path for every two vertices u and v, the graph G2 need not contain two
internally disjoint proper u − v paths. (See Exercises 5.40 and 5.41.)
Let G be a connected graph of diameter d ≥ 2 and let k be an integer with 2 < k < d. If Gk is a distance-
labeled graph and e is an edge labeled j where 1 < j ≤ k, then e is necessarily adjacent to an edge labeled i
for every integer i with 1 ≤ i < j (see Exercise 5.42). This property possessed by distance-labeled graphs
Gk suggests other concepts.
Let G be a connected graph. By a proper edge labeling of G we mean a labeling of the edges of G from
the set {1, 2, …, k} for some positive integer k such that no two adjacent edges are labeled the same.
We now turn our attention to proper edge labelings of trees. A proper edge labeling of a tree T from the
elements of the set {1, 2, …, k} is called a Grundy labeling if the labeling has the property that whenever
an edge e is labeled j where 1 < j ≤ k, then e is adjacent to an edge labeled i for every integer i with 1 ≤ i <
j. Grundy labelings are named for Patrick Michael Grundy (1917–1959), whose interests included
combinatorial games. The maximum positive integer k for which a tree T has a Grundy labeling from the
set {1, 2, …, k} is called the Grundy index Γ′(T) of T. For every tree T, the Grundy index Γ′(T) exists and
Δ(T) ≤ Γ′(T) ≤ 2Δ(T) − 1. (See Exercise 5.43.) For example, Γ′(T) = 4 for the tree T of Figure 5.15.

By an achromatic edge labeling of a tree T we mean a proper edge labeling of T from the elements of the
set S = {1, 2, …, k} for some positive integer k such that for every two distinct elements i and j of S, there
exist adjacent edges labeled i and j. The maximum k for which T has an achromatic edge labeling is
called the achromatic index of T and is denoted by ′(T). For example, ψ′(P9 ) = 4 (see Figure 5.16).
 SCRIPT
Intro clip
Graph Theory
Hello everyone! Today’s topic is Connectivity. So please stay tuned and keep on watching!
Liezel: Greetings. I am Liezel Aquino and I will be discussing *name ng chapter 5.1, 5.2* and also the
half of *name ng 5.3
--- discussion proper---
Esperanza: Good day! I am Ma. Esperanza Garcia and my discussion right now up until the end of this
chapter includes the Menger’s Theorem as well as the Power edges and labelings. So what are we waiting
for? Let’s start!
---discussion of Espie—
Menger’s Theorem:
Karl Menger


Born in Vienna, Austria on January 13, 1902.

Developed a talent for mathematics and physics.

Entered University of Vienna in 1920.

The following year he attended a lecture by Hans
Hahn on Neueres über den Kurvenbegriff
(What’s new concerning the concept of a curve)
and Menger’s interests turned towards
mathematics.
 Menger became quite ill while a student and his
studies were interrupted. During this time both of
his parents died. He eventually returned to the
university where he completed his studies under
Hahn in 1924.
 “Zur allgemeinen Kurventheorie” (Menger’s
theorem)
 In the spring of 1930, Karl Menger traveled to
Budapest and met many Hungarian
mathematicians, including Dénes König.
 Menger was considered one of the leading mathematicians of the 20th century.
 He died on October 5, 1985.
As we mentioned above, Theorem 5.16 is referred to as Menger’s Theorem, named for Karl Menger who
was born in Vienna, Austria on January 13, 1902. Menger developed a talent for mathematics and physics
at an early age and entered the University of Vienna in 1920 to study physics. The following year he
attended a lecture by Hans Hahn on Neueres über den Kurvenbegriff (What’s new concerning the concept
of a curve) and Menger’s interests turned towards mathematics. In the lecture it was mentioned that there
was no satisfactory definition of a curve (at that time), despite unsuccessful attempts to do so by a number
of distinguished mathematicians, including Georg Cantor, Camille Jordan and Giuseppe Peano. Some
mathematicians including Felix Hausdorff and Ludwig Bieberbach felt that it was unlikely that this
problem would ever be solved. Despite being an undergraduate with limited mathematical background,
Menger solved the problem and presented his solution to Hahn. This led Menger to work on curve and
dimension theory.
Menger became quite ill while a student and his studies were interrupted. During this time both of his
parents died. He eventually returned to the university where he completed his studies under Hahn in 1924.
The next year Menger went to Amsterdam to work with Luitzen Brouwer and Menger broadened his
mathematical interests. In 1927 Menger returned to the University of Vienna to accept the position of
Chair of Geometry. It was during that year that he published the paper “Zur allgemeinen Kurventheorie”
(Menger’s theorem). Menger himself referred to this result as the “n-arc theorem” and proved it as a
lemma for a theorem in curve theory.
In the spring of 1930, Karl Menger traveled to Budapest and met many Hungarian mathematicians,
including Dénes König. Menger had read some of König’s papers. During his visit, Menger learned that
König was working on a book that would contain what was known at that time about graph theory.
Menger was pleased to hear this and mentioned his theorem to König, which had only been published
three years earlier. König was not aware of Menger’s work and, in fact, didn’t believe that the theorem
was true. Indeed, the very evening of their meeting, König set out to construct a counterexample. When
the two met again the next day, König greeted Menger with “A sleepless night!”. König then asked
Menger to describe his proof, which he did. After that, König said that he would add a final section to his
book on the theorem, which he did. This was a major reason why Menger’s theorem became so widely
known among those interested in graph theory.
Because of the political situation in Austria in 1938, Menger left for a position in the United States at the
University of Notre Dame. While there, his work emphasized geometry, which didn’t have the impact of
his earlier work since geometry was not a subject of great interest to many, especially in the United
States. Menger went to the Illinois Institute of Technology in 1948 and spent the rest of his life in the
Chicago area. Menger was considered one of the leading mathematicians of the 20th century. He died on
October 5, 1985. With the aid of Menger’s theorem, Hassler Whitney was able to present a
characterization of k-connected graphs.

There are many theorems in mathematics which state that the minimum number of elements in some set
equals the maximum number of elements in some other set. In today’s lesson, we are going to prove
menger’s theorem, so let us read this quickly.
Theorem 5.16 (Menger′s Theorem) Let u and v be nonadjacent vertices in a graph G. The minimum
number of vertices in a u − v separating set equals the maximum number of internally disjoint u − v paths
in G.
Slide ppt 1:
Now we get to the real deal, the proving. As it turns out it is going to be a proof by induction on the size
of the graph. So the basis steps of the induction of the proof is to prove the result is true for all graphs that
have a size of zero or what we call as empty graphs then Menger’s theorem is true.
For induction step, we assume that menger’s theorem is true for all graphs of size less than m where m is
just some natural number and it is greater than 0. Let G be graph of size m. Let u and v be two
nonadjacent vertices of G. (Remember that Menger’s theorem is about any pair of non adjacent vertices in
a graph so next we’ll just take two non adjacent vertices and we’ll call them u and v). Suppose there are k
vertices in a minimum u-v separating set. Prove: Maximum number of internally disjoint u-v paths is K.
Let’s mention that we know the result is true for K=0 and for K=1. So again if the minimum number of
vertices we need to delete is 1, then there is certainly a maximum of just 1 internally disjoint uv paths.
There may be more than a single path connecting u and v but they cannot be possibly be disjoint.
So for the rest of the proof, we can proceed that assuming K is greater than or equal to 2. One last thing is
that, Certainly there are not more than K internally disjoint u-v paths. Then we will established k as the
maximum proving the theorem.
Slide ppt 2:
So for case 1. This is going to be a proof by cases. There exists a minimum u − v separating set W in G
containing a vertex x that is adjacent to both u and v.

Remember again that x is a vertex while u and v are two nonadjacent vertices that can be separated by
deleting a minimum of k vertices, thus our separating set w must have k vertices. In this case, we are
going to use the induction hypothesis.
Consider the graph G minus vertex x has a size less than m. And once we delete X, we get rid of at least
2 edges.
W – {x} is a minimum u-v separating set in G-x.
The cardinality of |W – {x}| must be k-1 because the cardinality of W is K so once we delete an element
of W that leavers K-1 elements.
That what will we do? WE can apply our induction hypothesis to the graph G-x which tells us since there
is a maximum of k-1 internally disjoint u-v paths in G-x. And then we get the final point of case 1. These
k-1 internally disjoint u-v paths of G-x, together with P= (u,x,v), make up K internally disjoint u-v paths
in G. Remember that vertex X is the vertex of our minimum u-v separating set W in G which is adjacent
to both u and v vertices. So that gives us one UV path that is not in G-x since g-x does not contain the
vertex x so we just include that path along with the K-1 internally disjoint u-v paths that we know are in
g-x and that guarantees us K internally disjoint u-v paths in G and that proves the theorem in case 1.
On to the Case 2. There exists a minimum u − v separating set W in G containing a vertex in W that is not
adjacent to u and a vertex in W that is not adjacent to v.
Let’s say we have W= {w1, w2, w3…wk} remember the minimum uv separating sets must contain K
vertices. Then. What we’re going to do is we are going to split the graph G into two sub graph 1
containing the vertex u, and one containing the vertex v.

Then with the use of some arguments, we are going to use induction hypothesis to both sub-graphs and
stitch them together. So let G u be the subgraph of G consisting of all u-w i paths where only w1 € W in
each path. let Gu be the subgraph of G consisting of all u-w i paths where only w1 € W in each path. And
let Gv be the subgraph of G consisting of all v-w i paths where only w1 € W in each path. So here is the
diagram that’s going to help us walk through the rest of this case.

So our original graph kind of look like this right, and then what we have just done is that we split the
graph in two:
This is what Gu looks like:

And this is what Gv looks like:

The path stops as soon as it gets to the set W. Suppose we slightly change the Subgraph Gu. Let us call it
G prime u, in this new version we add a new vertex called v’ and we join this vertex by an edge to every
vertex in the minimum uv separating set w. So the white lines are path while green lines are edges.

Now, we can apply our induction hypothesis to u and v’. So we can conclude that the size of G’u is less
than m. So finally, the cardinality of a minimum u-v prime separating set in this graph is K. The induction
hypothesis tells us that there exists a maximum of k internally disjoint u-v’ paths in G’ u, each one
consisting of a u-wi path Pi, followed by wiv’.

So by adding the vertex v’, we were able to apply our induction hypothesis and we basically used all
those paths through the vertices in W. Now I hope you understand that we could apply the same exact
arguments to our other subgraph Gv. By turning it to G’v by adding verticea new vertex u’ with the
same reason, it will lead us to the conclusion of there exists a maximum of k internally disjoint u’-v paths
in G’v, each one consisting of a wi – v path Qi, followed by u’wi.

Now that we get to our concluding statement: The k paths consisting of P i followed by Qi for 1 ≤i≤k are k
internally disjoint paths in G.
Let’s go ahead and move on to our final case. Case 3 states as, For each minimum u − v separating set W
in G, either every vertex of W is adjacent to u and not adjacent to v or every vertex of W is adjacent to v
and not adjacent to u.

Let’s start our argument with a very special type of path a UV geodisc. So we’ll say, Let P = (u, x, y…v)
be a u-v geodisc. That means it’s a shortest u-v path. We know P has at least two internal vertices which
we’re calling x and y. So let’s call the edge that joins x and y as e.
e=xy.
Then consider the graph G-e, so we deleted that edge joining x and y. If we delete the edge joining x and
y and we consider the graph G-e, certainly, Every minimum u-v separating set in G-e contains at least k-1
vertices.
Let us use a contradiction proof so let’s say, SFC or suppose for contradiction that every minimum u-v
separating set in G-e contains k-1 vertices.
Let S = {S1, S2, S3, … Sk-1} be a minimum u-v separating set in G-e . if once we delete the edge e from G,
we need to delete a minimum of these vertices to disconnect U and V then certainly in our original graph
G if we delete vertex X and vertices S, that would be a minimum way to separate u and v.
Then S U(union) with {x} is a minimum u-v separating set in G.
Thus all vertices of S are adjacent to U.
In the same way of deleting x, gets rid of the edge E and then we can separate u and v by deleting the
vertices of S, we could also delete the vertex y which has the effect of deleting its incident edge E.
We could get another minimum uv separating set in G such as
Then S U(union) with {y} is a minimum u-v separating set in G but since all the vertices of S are adjacent
to U and Y on the other hand is a minimum u-v separating set of G, that means that y is adjacent to the
vertex U as well but THAT IS A CONTRADICTION because remember our path P which was
introduced a while ago is a geodisc which is a shortest uv path. So if vertex y is adjacent to the vertex u,
then there would be a shorter path from u to v.
Thus, every minimum u-v separating set in G-e contains k vertices.
By the induction hypothesis, there exist a maximum of K internally disjoint u-v paths in G-e.

Power Edges and Labelings:

Now, it is time to move on to the power edges and labelings. But before that, we need to consider the
following figures.
Unlabled graph – from the term itself, no label at all.
Edge-labeled graph – the labels of the graphs are on the edges.
Which is the opposite of the vertex-labeled graph which the labeled ones are the vertices.
Vertex-colored graph - the vertices of the graphs are colored.
Which is opposite to the edge-colored graphs because the colored ones are the edges.
Lastly we have the vertex and colored graphs, meaning both of the vertices and edges are colored.
In particular, for every connected graph G of order n ≥ 3 and every two distinct vertices u and v of G,
there exist in G2 two internally disjoint u − v paths P and P′. Of course, each edge of P and P′ belongs
either to G or to G2 − E(G).
Distance-labeled graph – where in labelings occur when the label of each edge of G^2 that belongs to G
with the label 1 and label each edge of G2 not belonging to G with the label 2. In general, an edge uv of
Gk is labeled i (1 ≤ i ≤ k) if dG(u, v) = i.
A path P in Gk is called proper if every two adjacent edges in P have different labels.
However when it comes to proper edge labeling of a tree T from the elements of the set {1, 2, …, k} is
called a Grundy labeling if the labeling has the property that whenever an edge e is labeled j where 1 < j ≤
k, then e is adjacent to an edge labeled i for every integer i with 1 ≤ i < j. Grundy labelings are named for
Patrick Michael Grundy (1917–1959), whose interests included combinatorial games. The maximum
positive integer k for which a tree T has a Grundy labeling from the set {1, 2, …, k} is called the Grundy
index Γ′(T) of T.
For example this one, this has a grundy index of 4 based on the maximum positive integer it has. Proper
labeling occurs here because the are no adjacent numbers that are the same.
Another thing we have here is the achromatic edge labeling of a tree where it also used a proper edge
labeling of T from the elements of the set S= {1,2,…,k}for some positive integer k such that for every
two distinct elements i and j of S, there exist adjacent edges labeled i and j. Again, the maximum k that
you’ll see is called an achromatic index of T denoted as ‘(T).
This is an example of achromatic index 9 with P9.