You are on page 1of 4

MATH 273: Homework 13

Erik Brown
December 10, 2018

Challenge 1
Let us begin by finding ex(n,P2 ). To do this note that any vertex in our graph
cannot have degree 2 or more, since having degree 2 implies that there are two
edges who share a vertex, and that gives us a P2 . Therefore, the best we can
do is have every vertex of degree 1 if possible, and of degree 0 otherwise. If
n is even, we have enough vertices to form a perfect matching, which would
mean that every vertex is of degree 1, and since that is the maximum degree
for every vertex, it is also the maximum edges since you increase the degree of
two vertices every time an edge is added. This gives n/2 edges since each edge
is shared by two nodes. When n is odd, we cannot have a perfect matching,
so we will match n-1 of them, and leave 1 with degree 0. This gives number
of edges equal to the floor of n/2, which is no greater than n/2, so we can say
that ex(n,P2 ) is n/2 . For ex(n,P3 ), we can now have vertices of any degree,
since we could have the case of a star graph which has degree n-1, but no P3 .
However, with the exception of K3 , no two vertices of degree higher than 1 can
share an edge. The exception of K3 is because the start and end of our path
are the same, so therefore, we do not have the P3 subgraph, where if there were
any other edges connected to this structure, we could find our P3 . Therefore,
our maximum number of edges is to make the maximum number of K3 ’s and
then match two vertices if there are 2 left. Since K3 adds 3 edges for each 3
nodes, if the number of nodes is a multiple of 3, we have n edges. For the case
when the number of nodes not divisible by 3, but is even, we can we can create
K3 for all nodes except 2, and the last two we can add an edge between, so we
then have n-1 edges. For the case when n is not divisible by 3 or two, we also
have n-1 since the extra vertex is simply isolated. Therefore, the ex(n,P3 ) is n
for n divisible by 3, and n-1 otherwise.

Challenge 2
Note that in this structure, the average degree is of size 2, so this is a bound on
our extremal number of edges. To get a graph without this substructure, but
with degree of 2 almost everywhere, let us do what we did in challenge 1. This
gives n for divisible by 3, and n-1 for not divisible by 3.

1
Challenge 3
Note that we have an upper bound with 18, since R(3,6)=18. Let us denote
color 1 as red, color 2 as blue, and color 3 as green. Then R(3,6) is the largest
n such that any edge coloring of Kn using blue and red will have either a red
triangle, or a blue K6 . Now if we consider R(3,3,3), we know with 18 we either
get a red triangle or a blue K6 . A red triangle satisfies R(3,3,3) since this is the
maximum complete graph such that any edge coloring using red blue and green
guarantees a monochromatic triangle, so if we have a red one, we are done. If
not, we have a K6 that does not use red, so it must use blue and green. But we
know R(3,3)=6, so in this case we are still guaranteed a blue or green triangle.
Since 18 is an upper bound. Let us try to show true for 17. To do so, let us
consider an arbitrary vertex v. since v has 16 incident edges, we know that there
are at least 6 edges of the same color. Without loss on generality, let us say
that these are red edges. Now, if there are any red edges between the vertices
in the neighborhood of v, we have a red triangle, so we have a K6 of only blue
and green edges. But since we know R(3,3)=6, we know that this K6 has a blue
or green triangle, so therefore we know that 17 is an upper bound for R(3,3,3).
To show that R(3,3,3) is equal to 17, let us give a 3 edge coloring of K1 6 such
that there exists no monochromatic triangle. This is shown below, and since it
can be done, we know that R(3,3,3) is greater than 16, and less than or equal
to 17, so it is equal to 17.

Challenge 4
Let us begin by showing K1 0 is an upper bound. Once again denote this problem
as containing a red triangle or a blue K4 . Begin by choosing an arbitrary vertex,
and noting there are 9 incident edges. This means there are at least 4 red, or
at least 6 blue. When there are at least 4 red, if any of the edges between the
endpoints are also red, we end with a red triangle, which we can’t have. But this
means that every edge from the endpoints must be blue, which creates a blue
K4 . When there aren’t 4 red edges, there are 6 blue. For this case, we have our
K6 , and since we know that K6 has either a red or blue triangle, we are done
since the red triangle is forbidden, and the blue triangle with the blue edges
to our chosen vertex gives a K4 . To show that 9 is also an upper bound, note
that we cannot ensure that there are at least 4 red or 6 blue edges, since there
could be 3 red and 5 blue. However, we can find a vertex that does not have
this structure, since if every vertex had this same coloring of its neighborhood,
9 vertices would see 3 red edges, which means that there are 13.5 red edges in
the graph, which cannot happen. Therefore, we can pick a vertex which has
either 4 red of 6 blue edges, and apply the same as above. To show that 9 is
the bound, let us color K8 such that there are no red triangles or blue K4 as
shown below. Since we can color K8 , but not K9 , 9 is our answer.

2
Challenge 5
To show that finite paths are well quasi ordered under the subgraph relation,
we need to show that the subgraph relation is reflexive,transitive, and that for
any infinite sequence of elements, there exists two elements where element 1 is
a subgraph of element 2, and element 2 comes later in the sequence. For paths,
we know the subgraph relation is reflexive because every graph is a subgraph
of itself. For transitivity, note that Pn−k is a subgraph of Pn for all k from all
k between 0 and n. Now we need to prove that for arbitrary paths x,y,and z,
if y is a subgraph of x, and z is a subgraph of y, z is a subgraph of x. To do
this, let x be Pn . Then if y is a subgraph of x, y is Pn−k for some k between 0
and n. Since z is a subgraph of y, it is of the form Pn−k−j for some j between
0 and n-k. But since k-j is a constant, we can call it r, and now we have z of
the form Pn−r , which is a subgraph of Pn , proving that z is a subgraph of x if
y is a subgraph of x and z is a subgraph of y, which is our transitive property.
To show that finite paths are well ordered, let us note that there is no infinitely
strictly decreasing sequence since we are only discussing finite paths, so after
removing all vertices there will be no more paths to continue the sequence.
Now consider an arbitrary sequence that is not strictly decreasing. This means
it must be either not increasing anywhere, or increasing somewhere. When it
is not increasing anywhere, since we know it can’t be decreasing everywhere, so
therefore, there are two paths that are the same, and are thus related due to the
reflexivity of the subgraph relation. For the case when our paths are increasing,
there exists a path of length longer that a previous one, since that is how paths
increase. But every path is a subgraph of every path longer that or equal to
itself, so therefore the previous graph is related to the longer one, which is later
in the sequence since it is increasing. Therefore, for both of these cases we have
found our pair that is well ordered, so we know that the subgraph relation on
finite paths is well quasi ordered. To show finite trees are not well quasi ordered
under the subgraph relation, let us construct an infinite anti chain. To do this,
let us construct an infinite sequence of trees that are not related to each other.
Begin with vertices a,b,c,d,e, and f with edges ab,bc,de,and ef. Now add a path
of length n between vertex b and vertex e, and let the sequence be the increasing
length of this path. That is the first tree is with edge eb added, the next adds
vertex g and edges bg and ge, and the path gets infinitely larger. None of these
graphs are related to any other, so there cannot be a well ordered pair in the
sequence, therefore proving that the subgraph relation on finite trees is not well
quasi ordered.

Challenge 6
To show this note that two bags are adjacent only when the subtrees in the bag
intersect, which means every intersection of adjacent bags is non empty. This
also means that removing the intersection removed the overlapping part of the
two trees represented by the bags. But since we remove the intersection, the

3
two subtrees are no longer connected via this edge in the tree decomposition.
Since the tree decomposition is a tree, this means there is only 1 path to each
bag, so there is only 1 path between the subtrees in the graph. But since we
removed that path, and we know there is not another one, we have disconnected
the graph, which means we have a vertex cutset.

You might also like