2.

DIRECTED GRAPHS

In some cases, it may be useful to associate a direction to the edges of a graph. As such,
instead of having an {x, y} edge linking the x vertex and the y vertex, we might have an
edge directed from x to y or from y to x. We call this an “arc” and we use the notation
(x,y) to represent an arc leading from x to y.
Sometimes we may be dealing with “mixed” graphs, when some edges, but not all of
them, have a direction.

In Chapter 4, Bonvin draws a mixed graph when he tries to reproduce the configuration
of the treasure hunt location. The arcs represent the path that would make it possible to
gather the clues left by Mr. Grumbacker.

For his part, Manori uses directed graphs in Chapter 8 to try and push back the Courtel
family’s wake-up time on Thursday morning.

All the concepts defined above for non-directed graphs can also be extended to directed
graphs. For instance, we get the following definitions.

Definition
A “path” in a directed graph is a succession of arcs that allows you to move from one
vertex to another.

Example: in Chapter 8, Manori draws a path linking E1 to D6.

Definition
A directed graph is “strongly connected” if there is a path from x toward y and from y
toward x for every x, y pair of vertices in the graph.

Example: the directed graph drawn by Manori to push back the Courtels’ wake-up time is
not strongly connected because, for instance, there is no path leading from G1 to E1. To
make it strongly connected, all we would have to do is add an arc from D6 toward E1.

Definition
A “circuit” in a directed graph is a path whose first extremity coincides with its final
extremity.

Example: the graph below has a circuit containing vertices a, b and c.

Paths and circuits can be elementary or simple, just like chains and cycles. They are
elementary if they don’t touch the same vertex twice, and they are simple if they don’t go
over the same arc twice.

3. DEGREES OF VERTICES

In the book’s first chapter, Manori explains to Courtel that

“When you’re trying to draw a graph, you can always count the number of edges
touching each vertex.”

Definition
The number of edges that touch a vertex determine its “degree.”

So for example, in the graph below, vertices a, c and e each have a degree of 2, while
vertices b, d and f each have a degree of 4.

The property that permits Manori to demonstrate that the conference organizers can’t
create groups within the constraints they have established is the following one.
Property
The sum of the degrees of each vertex in a graph is always an even number.

Proof
When you add together the degrees of the vertices, each edge is counted twice. Each
edge {x, y} will be counted first when we consider the degree of vertex x, and a
second time when we add the degree of y. This sum is therefore double the number of
edges in the graph, and for that reason, it must be an even number.

In the example above, we have three 2-degree vertices and three 4-degree vertices, which
gives a total of 32 + 34 = 18, or the double of 9, which is the number of edges in the
graph.

This is how Manori succeeds in convincing Sébastien that the groupings the organizers
want to create cannot exist. The graph he uses includes one vertex per COPS participant,
and one edge between two participants when those participants have worked together in
the past. Requiring that each person be in a group of 15 including seven people they have
already worked with is equivalent to trying to build a graph with 15 vertices, each with a
degree of 7. Given that the total of the degrees in this hoped-for graph would be 715 =
105, which is an odd number, the graph the organizers want to build cannot exist.

Other applications:
o It is just as impossible to connect 15 computers with cables such that each computer
has a direct link with exactly seven other computers. In this case the vertices are
computers and the edges represent the cable links between the computers.
o It is also impossible to draw 15 line segments on the plane such that each segment has
a non-empty intersection with exactly seven other segments. The vertices here
correspond with segments and the edges with intersections between segments.
However, if we want six segments that each touch exactly three other segments, that
is possible, for example using the six segments below:

which can be represented with this graph:

Other very simple properties dealing with vertex degrees can be demonstrated. Here is
another example.

Property
In every graph, there are at least two vertices of the same degree.

Proof
Consider a given graph with x number of vertices. The degree of each vertex is
therefore a number chosen from the set {0, 1, …, x-1} since each vertex can only be
linked at most to x-1 other vertices.
If all the degrees are different, given that {0, 1, …, x-1} contains x values and that the
graph contains x vertices, there must necessarily exist a vertex with a degree of y for
each y in the set {0, 1, …, x-1}.
Hence, there must be a 0-degree vertex that is not linked with any other vertex, and
there must also exist a (x-1)-degree vertex which is linked to all the other vertices.
But this is in fact a contradiction, because the (x-1)-degree vertex cannot be linked
with the 0-degree vertex (which is not linked to any other vertex).

Application:
Let’s imagine that the COPS organizers decide to change the rules for creating their work
groups. They could, for instance, decide that when two people are in a same group, they
know a different number of people in the group. Manori would then have been easily able
to prove that such a grouping isn’t possible, since the graph associated with each group
would have to include vertices that are each of different degree, which is impossible, as
determined by the property set out above.