0:07

Today, we're going to talk about connectivity in networks.

First, we're going to talk about connectivity in undirected graphs.

Those are the ones where the edges don't have a direction.

An undirected graph is said to be connected if for every pair of nodes,

there is a path between the two nodes.

We can use the function is_connected in

network X and give it the undirected graph as input,

and it will tell us whether the graph is connected or not.

In this case, this example,

this graph is connected so it says, true, it is connected.

However, if we were to remove a few of the edges, for example,

if we remove A-G, A-N,

and J-O, then the graph will become disconnected, as you can see.

Now, we have these sort of

three communities such that if you're in a particular community,

you cannot find a path to a node in a different community,

or in a different set of nodes.

Let's try to get at this idea of communities in a more precise way.

We're going to refer to these communities as connected components.

So let me give you a definition of what a connected component is.

A connected component is a subset of nodes

such that there are two conditions this set of nodes satisfy.

First, every node in the subset has to have a path to every other node in the subset.

That's condition number one.

Condition number two would say that no other node outside of

the subset has a path to any node inside the subset.

So condition two kind of makes sure that

you get all the nodes that you could possibly can

so that every node in this subset has a connection to every other node in the subset,

not a subset of the subset that you could potentially get.

So let's see this through examples to make it more clear.

Is the subset of nodes E, A,

G, F a connected component?

So first, let's find these nodes and so here they are.

This is the subset of nodes we're referring to.

And we can clearly see that this is not a connected component because the nodes,

for example, A and F,

cannot reach each other.

There is no path going from A to F,

so condition number one fails.

Okay, let's look at another example.

Is the subset of nodes N,

O, K a connected component?

All right, these are the nodes.

Now in this case, condition number one is actually met.

There is a path from any node in N,

O, K to any other node in N, O,

K. For example, if you wanted to find a path from N to K,

you would go N-O-K.

However, condition number two fails

because there are other nodes that can actually reach nodes in the subset.

For example, L can actually reach N,

O and K, there is a path from L to all three of those nodes.

So therefore, this is not a connected component because condition number two fails.

So if you think about it,

the only things that satisfy the definition of

connected component are the things that we originally started with.

These three communities such that every node is

connected within and no nodes are connected across.

And these are the three connected components in this particular graph.

You can use network X to find the connected components of

an undirected graph by using the function number_connected_components and give it,

the graph, its input and it would tell you how many.

It has, in this case, three.

And you can also ask for which ones

are the actual components by using the function connected_components(G),

and here we tell you which are

the subsets of nodes and which nodes belong to each one of the components.

What you can also do is use the function node_connected_component

with the input G of the graph itself but also a particular node as input,

and this would tell you which connected component this particular node belongs to.

So in this case, for example,

if you ask for the component in which M belongs,

then it would tell you that it belongs to a component K, L, M, N, O.

So these are useful things to use in network X.

Now, let's talk about connectivity in directed graphs.

Remember, directed graphs are those that have edges where the edges have direction,

so there is a source and a destination.

And so, here is an example of a directed graph.

We would like to come up with definitions of

connected and connected components that apply to directed graphs,

but because paths have a different definition in

directed graphs than they do in undirected graphs,

then we need to adjust our definitions accordingly.

And actually, what happens is that we will have

two types of definitions for each one of the two concepts.

So let's start with the concept of connected.

The first type is we're going to say that a direct graph is

strongly connected if for every pair of nodes,

say U and V, there is a directed path that goes from U

to V and another directed path that goes from V to U.

That, if a directed graph has a has a property,

then we say it's strongly connected.

We can use the function is_strongly_connected in

network X to ask whether this particular directed graph, G,

is strongly connected and it would say false because if you look carefully,

there is no path that goes from A to H, for example.

There are many other examples of pairs of

nodes for which there is no path, where here is one,

there is no path from A to H, so therefore,

this graph is not strongly connected.

The second definition for connected is weakly connected.

And the way weakly connected works is that the first thing you do is you

replace all the directed edges and you make them undirected.

So every edge, you could sort of ignore

the direction and you make him into a undirected edge,

and now this graph becomes undirected.

And now, you ask the question that you already applied to undirected graphs,

is this graph connected or not?

And if it is, then we say that the original directed graph is weakly connected.

So in this case, if we use the function is_weakly_connected,

network X would say yes,

this graph is weakly connected because once you turn it into an undirected graph,

this undirected graph is connected.

All right, now let's talk about the connected component definition.

Again, we're going to have two types.

The first one is strongly connected components

and the definition mirrors the definition for undirected.

It's just that now, you have to find paths that are directed,

so it would have the two conditions for a subset of nodes to be

connected component are that every node in the subset has a directed path to

every other node in this subset and that no node outside

the subset has a directed path to and from every node inside the subset.

And so in this case,

what are the strongly connected components of this graph?

It's actually not that easy to tell visually.

It's a little tricky, though you could try to

pause the video and try and see if you can find what they are.

But if we use the function strongly_connected_components,

it will tell us what they are.

And in this case, it turns out that these are the strongly connected components.

And so, as you can see,

for example, you find that N, node M,

and node L are in different components

and that's because even though you can go from L to M,

you cannot go from M to L, right?

And same thing with, for example,

H, I are sort of in their own component,

and that's because while they can reach other nodes like G and J,

none of those nodes seem that they can reach them.

And so, these live in their own separate, strongly connected component.

And of course, we would have

the weakly connected component version which works in the same way that it did before.

So first, we would make all the directed edges undirected,

and then we would find the connected components in the new undirected graph.

Now, because this graph is weakly connected,

that means that when you make all the direct edges undirected,

it becomes a connected graph.

Then this particular graph has only one weakly connected component,

which is the whole graph.

And so, to summarize,

we talked about undirected graphs and directed graphs and we were

talking about a couple of definitions of connectivity.

So for underactive graphs,

we said that an undirected graph is connected if for every pair of nodes,

there is a path between them.

And we talked about connected components and we said that we could

use the function connected_components to find these connected components,

so here's an example.

Now, for the directed case,

we had two types of definitions,

the strong and the weak.

For the strongly connected,

we said that our graph is strongly connected if every pair of nodes,

they have a directed path from one node to the other and from the other node to the one,

and you could use the function strongly_connected_components

to find what these components were.

And both of these had the corresponding weak definitions,

so weakly connected and weakly connected components.

And the way that those work is by making the direct edges undirected,

and then applying the same definitions that we had from

undirected graphs to the new undirected graph that comes from the directed graph.

Thank you for watching and we'll see you next time.