Scribd
Upload
49 views2 pages

REMINDER: Adjacency Matrix N×N ( ) Is Called Invertible If Exist A

1. The document discusses properties of random walks on different graph structures, including the ring graph and star graph. 2. For the directed ring graph, a random walk starting at vertex 0 will not converge to a stationary distribution over any number of steps. However, a random walk starting with a uniform distribution may converge. 3. For the star graph, the adjacency matrix is not invertible and the distribution after one step of a uniform random walk is centered on the central vertex.

Uploaded by

Mushahid Hussain Nomee
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
49 views2 pages

REMINDER: Adjacency Matrix N×N ( ) Is Called Invertible If Exist A

1. The document discusses properties of random walks on different graph structures, including the ring graph and star graph. 2. For the directed ring graph, a random walk starting at vertex 0 will not converge to a stationary distribution over any number of steps. However, a random walk starting with a uniform distribution may converge. 3. For the star graph, the adjacency matrix is not invertible and the distribution after one step of a uniform random walk is centered on the central vertex.

Uploaded by

Mushahid Hussain Nomee
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

1.

The "Ring Graph"


Lets remind the Ring graph:

When
increment/decrement is done by MOD n.
a. Suppose that a random walk is executed on the graph vertices when the
distribution of vertices is centered on vertex number 0 (meaning vertex number
0 selected in probability 1 in the first step).
How would you look at the distribution of vertices after one step, after two,
after three?
What kind of distribution will it strive after t steps when t->∞.

b. How the distribution of the vertices will look after one step if you start the walk
in the uniform distribution? How would it look after t steps?

2. Looking on the directed ring graph:


In this graph there are n vertices numbered from 0 to n-1.
The edges groups is: 𝐸 = {(𝑖, 𝑖 + 1)|𝑖 ∈ [𝑛]} when increment is done by MOD n.
a. Show that if a random walk begins when the distribution of vertices is
centered on vertex 0 the walk will not converge for any number of steps.
Hint: think what would happen after 1, 2, n, n+1, …t steps.
b. Does the distribution of vertices converge if we begin with the uniform
distribution? If so, for which distribution will it converge?

3. The "Star Grapgh"


Let 𝑛 ∈ ℕ, n > 1.
We will define the graph: 𝑉 = {1, … , 𝑛}
The edges groups is defined as such:
For each 1 ≤ i ≤ n-1 there will be a edge (n,i), (i,n).
a. Is the Adjacency matrix an invertible matrix?
b. Let's start random walk which is distributed by the uniform distribution.
What will be the distribution of vertices after one step in the walk? How
will it look after two steps?
REMINDER: Adjacency matrix 𝐴 ∈ 𝑀 n×n(𝑹) is called invertible if exist a
𝑘 ∈ 𝑵 so for every i,j.

4. Let 𝐺 = (𝑉, 𝐸) strong connected directed graph (between each 2


vertices u,v there is a path from u to v and vice versa).
Let's look at the graph G' which is the same as G but in addition each vertex is added
a loop edge (to itself).
a. Prove that G' 's adjacency matrix is invertible.
Hint: Use the connection between the matrix's subjects and the
existence of k long routes.
Show the existence of a t ∈ ℕ which has a route between each pair of
vertices in G' that is t long.
5. Lets look at the following graph:

𝑉 = {1, … , 𝑛 + 1, 𝑛 + 2}

Where it's edges group is defined as such:

For each 1 ≤ i ≤ n-1 there will be an the edge (𝑛 + 1, 𝑖), (𝑖, 𝑛 + 1).

And in addition the edges: (𝑛 + 1, 𝑛 + 2), (n + 2, n + 2).

a. Is the Adjacency matrix an invertible matrix?


b. Find a stationary probability vector to the normalized adjacency
matrix (no need to find the exact value but rather the magnitude of
them. Meaning for each coordinate determine if it's value is 𝜃(1) or
𝜃(1/𝑛).

You might also like