You are on page 1of 9

Q.

Show that the following conditions are equivalent for graph 𝐺 of size at least 2:

• 𝐺 is connected and has no cut-vertex,

• any two vertices are on a cycle,

• any two edges are on a cycle,

• for any three vertices 𝑥, 𝑦 and 𝑧, there is an 𝑥, 𝑧-path containing 𝑦.

Sol : To show that the given conditions are equivalent for a graph 𝐺 of size at least 2, we'll prove that
each condition implies the others.

1. 𝐺 is connected and has no cut-vertex implies any two vertices are on a cycle:

- If 𝐺 is connected and has no cut-vertex, it means that there is a path between any pair of vertices
in 𝐺.

- Take two vertices 𝑥 and 𝑦. Since 𝐺 is connected, there exists a path between 𝑥 and 𝑦.

- Now, consider the path between 𝑥 and 𝑦. If it is a cycle, we are done. If it is not a cycle, we can
extend it by adding edges from 𝑥 to 𝑦 until we form a cycle. This extended path must exist because 𝐺
is connected and has no cut-vertex. Therefore, any two vertices are on a cycle.

2. Any two vertices are on a cycle implies 𝐺 is connected and has no cut-vertex:

- If any two vertices in 𝐺 are on a cycle, it means that there is a path between any pair of vertices in
𝐺 because you can always travel along the cycle to connect any two vertices.

- If 𝐺 had a cut-vertex, removing that vertex would disconnect the graph, which would mean there
are two vertices that are not connected by a path. This contradicts the assumption that any two
vertices are on a cycle. Therefore, 𝐺 must be connected and have no cut-vertex.

3. Any two edges are on a cycle implies any two vertices are on a cycle:

- Consider any two vertices 𝑥 and 𝑦 in 𝐺. We want to show that there exists a cycle containing 𝑥 and
𝑦.

- Let 𝑎 and 𝑏 be two arbitrary neighbors of 𝑥 and 𝑦, respectively. Since 𝑎 and 𝑏 are connected to 𝑥
and 𝑦, there exist edges (𝑥, 𝑎) and (𝑦, 𝑏) in 𝐺.

- By the condition that any two edges are on a cycle, there must be a cycle that includes both of
these edges. This cycle also includes 𝑥 and 𝑦, as they are endpoints of (𝑥, 𝑎) and (𝑦, 𝑏). Therefore,
any two vertices 𝑥 and 𝑦 are on a cycle.

4. For any three vertices 𝑥, 𝑦, and 𝑧, there is an 𝑥, 𝑧-path containing 𝑦 implies any two vertices are on
a cycle:
- Consider any two vertices 𝑥 and 𝑦 in 𝐺. We want to show that there exists a cycle containing 𝑥 and
𝑦.

- Take 𝑧 to be any third vertex that is not equal to 𝑥 or 𝑦. By the given condition, there exists an 𝑥,
𝑧-path containing 𝑦.

- This 𝑥, 𝑧-path, along with the edge from 𝑧 to 𝑦, forms a cycle containing 𝑥 and 𝑦. Therefore, any
two vertices 𝑥 and 𝑦 are on a cycle.

Now, we have shown that all four conditions are equivalent for a graph 𝐺 of size at least 2.

Q. Let 𝑉 (𝐺) = ∪𝑘 𝑖

=1𝑉𝑖 be a partition of the vertex set of a connected graph 𝐺 into 𝑘 ≥ 2

nonempty subsets such that 𝐺[𝑉𝑖] is connected. Prove that there are indices 1 ≤ 𝑖 <

𝑗 ≤ 𝑘 such that 𝐺 − 𝑉𝑖 and 𝐺 − 𝑉𝑗 are connected.

Sol: To prove that there exist indices 1 ≤ 𝑖 < 𝑗 ≤ 𝑘 such that 𝐺 − 𝑉𝑖 and 𝐺 − 𝑉𝑗 are connected, we can
use a proof by contradiction.

Assume, for the sake of contradiction, that for every pair of indices 1 ≤ 𝑖 < 𝑗 ≤ 𝑘, either 𝐺 − 𝑉𝑖 or 𝐺 −
𝑉𝑗 is not connected.

Let's denote 𝐺 − 𝑉𝑖 as 𝐺𝑖 (the graph obtained by removing 𝑉𝑖 from 𝐺), and 𝐺 − 𝑉𝑗 as 𝐺𝑗 (the graph
obtained by removing 𝑉𝑗 from 𝐺).

Now, since 𝐺 is connected, there must exist a path between any two vertices in 𝐺. This includes
vertices in 𝑉𝑖 and 𝑉𝑗, as well as vertices in 𝐺𝑖 and 𝐺𝑗.

Let's consider two cases:

Case 1: 𝑉𝑖 and 𝑉𝑗 are disjoint subsets.

In this case, there are no vertices in common between 𝑉𝑖 and 𝑉𝑗. Since 𝐺 is connected, there exists
at least one edge connecting a vertex in 𝑉𝑖 to a vertex in 𝑉𝑗. Let's call this edge (𝑢, 𝑣), where 𝑢 is in
𝑉𝑖 and 𝑣 is in 𝑉𝑗.
Now, consider the path in 𝐺 from 𝑢 to 𝑣. This path must go through vertices outside of both 𝑉𝑖 and
𝑉𝑗 because 𝐺 is connected. Let 𝑥 be the first vertex encountered on this path outside of 𝑉𝑖 and 𝑉𝑗.

Then, 𝑥 cannot belong to 𝑉𝑖 or 𝑉𝑗. Thus, 𝑥 is in 𝐺𝑖 and 𝑥 is also in 𝐺𝑗. This means that there is a path
between 𝑢 and 𝑣 in both 𝐺𝑖 and 𝐺𝑗, which contradicts our assumption that either 𝐺𝑖 or 𝐺𝑗 is not
connected for any pair 𝑖, 𝑗.

Case 2: 𝑉𝑖 and 𝑉𝑗 have at least one common vertex.

In this case, there are vertices that belong to both 𝑉𝑖 and 𝑉𝑗. Let 𝑢 be a vertex that is in both 𝑉𝑖 and
𝑉𝑗. Since 𝐺 is connected, there exists a path between any two vertices in 𝐺, including 𝑢.

Now, let's consider two cases:

Case 2.1: The removal of 𝑉𝑖 disconnects 𝑢 from the rest of 𝐺.

In this case, 𝑢 is not connected to any vertex outside of 𝑉𝑖 in 𝐺𝑖. However, since 𝑢 is in 𝑉𝑗 as well, 𝑢 is
connected to at least one vertex in 𝐺𝑗. Therefore, 𝐺𝑖 is not connected, which contradicts our
assumption.

Case 2.2: The removal of 𝑉𝑗 disconnects 𝑢 from the rest of 𝐺.

In this case, 𝑢 is not connected to any vertex outside of 𝑉𝑗 in 𝐺𝑗. However, since 𝑢 is in 𝑉𝑖 as well, 𝑢 is
connected to at least one vertex in 𝐺𝑖. Therefore, 𝐺𝑗 is not connected, which contradicts our
assumption.

In either case, we arrive at a contradiction. Therefore, our initial assumption that for every pair of
indices 1 ≤ 𝑖 < 𝑗 ≤ 𝑘, either 𝐺 − 𝑉𝑖 or 𝐺 − 𝑉𝑗 is not connected must be false. Hence, there must exist
indices 1 ≤ 𝑖 < 𝑗 ≤ 𝑘 such that both 𝐺 − 𝑉𝑖 and 𝐺 − 𝑉𝑗 are connected.

Q. Let 𝑎 and 𝑏 be vertices of a tree 𝑇 at maximal distance 𝑑(𝑎, 𝑏) = 2𝑟, and let 𝑐 be a

vertex on the unique 𝑎, 𝑏-path at distance 𝑟 from 𝑎 and 𝑏. Show that 𝑐 is the unique

vertex of 𝑇 with 𝑑(𝑐, 𝑥) ≤ 𝑟 for every 𝑥 ∈ 𝑇.

Sol: Let's prove that vertex 𝑐 is the unique vertex in tree 𝑇 with 𝑑(𝑐, 𝑥) ≤ 𝑟 for every 𝑥 ∈ 𝑇.
First, observe that by definition, the distance between 𝑎 and 𝑏 in 𝑇 is 2𝑟, which means there is a
unique 𝑎, 𝑏-path with a length of 2𝑟. Since 𝑐 is a vertex on this unique 𝑎, 𝑏-path at distance 𝑟 from
both 𝑎 and 𝑏, it follows that 𝑐 is indeed at distance 𝑟 from 𝑎 and 𝑏.

Now, let's assume for the sake of contradiction that there exists another vertex 𝑥 ≠ 𝑐 in 𝑇 such that
𝑑(𝑐, 𝑥) ≤ 𝑟. This means that there is a path from 𝑐 to 𝑥 in 𝑇 with a length of at most 𝑟.

Since 𝑐 is at distance 𝑟 from 𝑎 and 𝑏, it implies that there is a 𝑎, 𝑐-path and a 𝑏, 𝑐-path, both with
lengths less than or equal to 𝑟. Let 𝑝𝑎 be the 𝑎, 𝑐-path, and let 𝑝𝑏 be the 𝑏, 𝑐-path.

Now, let's consider the path from 𝑎 to 𝑏 in 𝑇. It must pass through 𝑐 because 𝑐 lies on the unique 𝑎,
𝑏-path. This path can be divided into two parts: 𝑎 to 𝑐 (which is path 𝑝𝑎) and 𝑐 to 𝑏 (which is path
𝑝𝑏).

The length of 𝑝𝑎 (the 𝑎, 𝑐-path) is less than or equal to 𝑟, and the length of 𝑝𝑏 (the 𝑐, 𝑏-path) is also
less than or equal to 𝑟. Therefore, the total length of the 𝑎, 𝑏-path passing through 𝑐 is at most 2𝑟.

However, we initially defined the distance between 𝑎 and 𝑏 in 𝑇 to be 2𝑟, which is precisely the
length of the unique 𝑎, 𝑏-path. This means that there cannot be any other path between 𝑎 and 𝑏
with a length less than 2𝑟. Therefore, it is not possible for 𝑥 ≠ 𝑐 to be at distance 𝑟 from both 𝑎 and 𝑏,
as it would require a path between 𝑎 and 𝑏 shorter than 2𝑟, which is not possible in a tree.

Hence, 𝑐 is indeed the unique vertex in 𝑇 with 𝑑(𝑐, 𝑥) ≤ 𝑟 for every 𝑥 ∈ 𝑇.

Q. Denote by 𝛼(𝐺) the maximum cardinality of a set of independent vertices in 𝐺. Prove

that if 𝐺 does not contain a cycle, then 𝛥(𝐺) ≤ 𝛼(𝐺) and deduce that 𝑒(𝐺) ≤ 12

𝑛𝛼(𝐺),

where 𝑛 = |𝑉 (𝐺)|.

Sol: Let's prove the given statements step by step:

1. First, we will prove that if 𝐺 does not contain a cycle, then 𝛥(𝐺) ≤ 𝛼(𝐺).

Suppose 𝐺 does not contain a cycle. We want to show that the maximum degree 𝛥(𝐺) is less than
or equal to the maximum cardinality of a set of independent vertices 𝛼(𝐺).
Consider a set 𝑆 of independent vertices in 𝐺 with maximum cardinality 𝛼(𝐺). Let 𝑣 be a vertex in 𝑆.
Since 𝑆 is independent, 𝑣 is not adjacent to any other vertex in 𝑆, which means that the degree of 𝑣
(the number of neighbors of 𝑣) is less than or equal to 𝛼(𝐺).

Now, let's consider the vertex 𝑢 in 𝐺 with the highest degree (i.e., the vertex 𝑢 with 𝛥(𝐺) =
degree(𝑢)). Since 𝑆 is independent, 𝑢 is not in 𝑆, and therefore, 𝑢 and 𝑣 are distinct vertices.

Because 𝑢 has the highest degree, it is adjacent to more vertices in 𝐺 than 𝑣 is. This means that
degree(𝑢) > degree(𝑣), or equivalently, 𝛥(𝐺) > 𝛼(𝐺).

However, this contradicts the fact that 𝛼(𝐺) is the maximum cardinality of a set of independent
vertices, and 𝑆 is such a set with cardinality 𝛼(𝐺).

Therefore, 𝛥(𝐺) cannot be greater than 𝛼(𝐺), and we have shown that 𝛥(𝐺) ≤ 𝛼(𝐺).

2. Next, we will deduce that 𝑒(𝐺) ≤ 12𝑛𝛼(𝐺), where 𝑛 = |𝑉(𝐺)|.

We know that 𝛥(𝐺) ≤ 𝛼(𝐺) from the previous step. By the handshaking lemma, the sum of the
degrees of all vertices in 𝐺 is equal to twice the number of edges, i.e., 2𝑒(𝐺). Therefore, we can write:

2𝑛𝛼(𝐺) ≥ 2𝑛𝛥(𝐺) ≥ 2𝑛(𝛼(𝐺)).

Dividing both sides by 2𝑛 (which is positive), we get:

𝛼(𝐺) ≥ 𝑒(𝐺) / 𝑛.

Rearranging the inequality, we obtain:

𝑒(𝐺) ≤ 𝑛𝛼(𝐺).

Now, multiplying both sides of this inequality by 12:


12𝑒(𝐺) ≤ 12𝑛𝛼(𝐺).

Therefore, we have deduced that 𝑒(𝐺) ≤ 12𝑛𝛼(𝐺).

Q. Fleury gave the following algorithm for finding an Euler circuit 𝑥1𝑥2 … 𝑥𝑛 in a graph 𝐺.

Pick 𝑥1 arbitrarily. Having chosen 𝑥1, 𝑥2, … , 𝑥𝑘, put 𝐺𝑘 = 𝐺 − {𝑥1𝑥2, 𝑥2𝑥3, … , 𝑥𝑘−1𝑥𝑘}. If

every edge incident with 𝑥𝑘 in 𝐺𝑘 is a cut-edge (in particular, if 𝑥𝑘 is an isolated vertex

of 𝐺𝑘), then terminate the algorithm. Otherwise, let 𝑥𝑘+1 be a neighbour of 𝑥𝑘 in 𝐺𝑘

such that 𝑥𝑘𝑥𝑘+1 is not a cut-edge of 𝐺𝑘.

Prove that if 𝐺 has an Euler circuit, then the trail 𝑥1𝑥2 … 𝑥𝑙 constructed by the algorithm

is an Euler circuit.

Sol : To prove that the trail 𝑥1𝑥2 … 𝑥𝑙 constructed by Fleury's algorithm is indeed an Euler circuit if 𝐺
has an Euler circuit, we need to show two things:

1. The trail 𝑥1𝑥2 … 𝑥𝑙 is a circuit, meaning it starts and ends at the same vertex 𝑥1.

2. The trail 𝑥1𝑥2 … 𝑥𝑙 covers every edge of 𝐺 exactly once.

Let's prove both of these points:

1. To show that the trail 𝑥1𝑥2 … 𝑥𝑙 is a circuit, we need to demonstrate that 𝑥1 = 𝑥𝑙.

Fleury's algorithm starts with an arbitrary vertex 𝑥1. At each step, it selects a neighbor 𝑥𝑘+1 of the
current vertex 𝑥𝑘 and moves to that vertex. The algorithm continues until it encounters a situation
where all edges incident with 𝑥𝑘 in 𝐺𝑘 are cut-edges, indicating that 𝑥𝑘 is an isolated vertex in 𝐺𝑘,
and the algorithm terminates. In this case, 𝑥𝑘 is the last vertex, which means 𝑥𝑙 = 𝑥𝑘.

Since 𝑥𝑘 is the last vertex in the trail and the algorithm starts at 𝑥1, we have 𝑥1 = 𝑥𝑙. Therefore, the
trail forms a circuit that starts and ends at 𝑥1.

2. Now, let's prove that the trail 𝑥1𝑥2 … 𝑥𝑙 covers every edge of 𝐺 exactly once.
Suppose 𝐺 has an Euler circuit, which means that there exists a circuit that covers every edge of 𝐺
exactly once.

Fleury's algorithm starts with 𝑥1 and, at each step, moves to a neighbor 𝑥𝑘+1 in 𝐺𝑘 such that
𝑥𝑘𝑥𝑘+1 is not a cut-edge of 𝐺𝑘. Since 𝐺 has an Euler circuit, all edges of 𝐺 are part of this circuit.

Now, consider the operation of the algorithm. It removes edges (𝑥1𝑥2, 𝑥2𝑥3, …, 𝑥𝑘−1𝑥𝑘) from 𝐺,
but it does not remove any edges that are part of the Euler circuit, as those edges are not cut-edges
of 𝐺𝑘.

Therefore, at each step, the algorithm only removes edges that are not part of the Euler circuit. It
continues to do so until it reaches a point where 𝑥𝑘 becomes an isolated vertex in 𝐺𝑘, which means
it has no more outgoing edges. At this point, the algorithm terminates.

Since the algorithm only removes edges that are not part of the Euler circuit and terminates when
𝑥𝑘 becomes isolated, it means that the trail 𝑥1𝑥2 … 𝑥𝑙 covers every edge of 𝐺 exactly once, including
the edges of the Euler circuit.

Therefore, we have shown that if 𝐺 has an Euler circuit, then the trail 𝑥1𝑥2 … 𝑥𝑙 constructed by
Fleury's algorithm is indeed an Euler circuit in 𝐺.

Q. Let 𝐺 be a planar graph of order at least 3, with degree sequence (𝑑𝑛𝑑𝑛−1 … 𝑑1). Show

that

𝛴𝑑𝑖≤6(6 − 𝑑𝑖) ≥ 𝛴𝑛 𝑖

=1(6 − 𝑑𝑖) ≥ 12.

Deduce that if 𝛿(𝐺) ≥ 5, then 𝐺 has at least 12 vertices of degree 5, and if 𝛿(𝐺) ≥ 4

the 𝐺 has at least 6 vertices of degree at most 5.

Sol: Let's break down the proof step by step:

1. First, we'll prove the inequality:

∑𝑑𝑖≤6(6 − 𝑑𝑖) ≥ ∑𝑛 𝑖=1(6 − 𝑑𝑖) ≥ 12.


The left side of the inequality represents the sum of (6 - 𝑑𝑖) over all vertices 𝑑𝑖 with degrees at
most 6 in the graph 𝐺. The right side of the inequality represents the sum of (6 - 𝑑𝑖) over all vertices
𝑑𝑖 in the graph 𝐺.

To prove this inequality, consider two cases:

Case 1: 𝑑𝑖 ≤ 6 (degrees at most 6).

In this case, ∑𝑛 𝑖=1(6 - 𝑑𝑖) includes the sum of (6 - 𝑑𝑖) for all vertices in the graph, including those
with degrees greater than 6. Therefore, ∑𝑛 𝑖=1(6 - 𝑑𝑖) is at least as large as ∑𝑑𝑖≤6(6 - 𝑑𝑖).

Case 2: 𝑑𝑖 > 6 (degrees greater than 6).

If 𝑑𝑖 is greater than 6, then (6 - 𝑑𝑖) is negative, and the sum (6 - 𝑑𝑖) over such vertices will
contribute negatively to ∑𝑛 𝑖=1(6 - 𝑑𝑖). Therefore, ∑𝑛 𝑖=1(6 - 𝑑𝑖) is still at least as large as ∑𝑑𝑖≤6(6 -
𝑑𝑖).

Combining both cases, we have established the inequality:

∑𝑑𝑖≤6(6 − 𝑑𝑖) ≥ ∑𝑛 𝑖=1(6 − 𝑑𝑖) ≥ 12.

2. Now, we'll deduce that if 𝛿(𝐺) ≥ 5, then 𝐺 has at least 12 vertices of degree 5:

If 𝛿(𝐺) ≥ 5, it means that the minimum degree in 𝐺 is at least 5. Let 𝑑𝑖 be the degree of a vertex 𝑣𝑖
in 𝐺. Since 𝑣𝑖 has degree at least 5, we have 𝑑𝑖 ≥ 5.

Using the inequality from step 1, we can write:

∑𝑑𝑖≤6(6 − 𝑑𝑖) ≥ 12.

Since 𝑑𝑖 ≥ 5 for all 𝑑𝑖 in the sum, we can rewrite it as:


∑𝑑𝑖=5,6(6 − 𝑑𝑖) ≥ 12.

In this sum, 𝑑𝑖 can only take the values 5 or 6. For each vertex 𝑣𝑖 with degree 5 or 6, the term (6 -
𝑑𝑖) will contribute either 1 or 0 to the sum.

To get a sum of at least 12, we need at least 12 vertices with degrees 5 or 6. Therefore, if 𝛿(𝐺) ≥ 5,
then 𝐺 has at least 12 vertices of degree 5.

3. Finally, we'll deduce that if 𝛿(𝐺) ≥ 4, then 𝐺 has at least 6 vertices of degree at most 5:

If 𝛿(𝐺) ≥ 4, it means that the minimum degree in 𝐺 is at least 4.

Using the inequality from step 1:

∑𝑑𝑖≤6(6 − 𝑑𝑖) ≥ 12.

Since 𝑑𝑖 is the degree of vertex 𝑣𝑖, we want to find a lower bound on the number of vertices 𝑣𝑖
with degrees 6 - 𝑑𝑖 ≤ 5.

Now, consider the terms in the sum with (6 - 𝑑𝑖) ≤ 5. These terms can only be 1 or 0. To get a sum
of at least 12, we need at least 12 terms equal to 1.

Therefore, 𝐺 must have at least 12 vertices with degrees 6 - 𝑑𝑖 ≤ 5.

However, some of these vertices might have degrees 6, so to ensure that we have at least 6 vertices
with degrees at most 5, we need 𝐺 to have at least 12 vertices with degrees 5 or less.

Thus, if 𝛿(𝐺) ≥ 4, then 𝐺 has at least 6 vertices of degree at most 5.

You might also like