Scribd
Upload
20 views3 pages

Random Walks On 1 and 2 Dimensions

This document proves that a random walk on 1 or 2 dimensions will return to the origin with probability 1. It defines a random walk as a series of steps in random directions. It shows that the expected number of returns to the origin is infinite for 1 dimension by calculating the probabilities of ending at the origin for walks of even and odd length. It then extends this to 2 dimensions by treating the walk as two independent 1D walks.

Uploaded by

sheckercazav
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
20 views3 pages

Random Walks On 1 and 2 Dimensions

This document proves that a random walk on 1 or 2 dimensions will return to the origin with probability 1. It defines a random walk as a series of steps in random directions. It shows that the expected number of returns to the origin is infinite for 1 dimension by calculating the probabilities of ending at the origin for walks of even and odd length. It then extends this to 2 dimensions by treating the walk as two independent 1D walks.

Uploaded by

sheckercazav
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/ 3

Random walks on 1 and 2 dimensions

We will prove a random walk on 1 or 2 dimensions will return to the origin at


probability 1.

Firstly, what is a random walk on k dimension?


1
Let X be a random variable such that ∀1 ≤i ≤ k P ( X =e i) =P ( X=−ei ) = where ei is
2k
the i'th vector of the normal basis for Rn . Let Sn denote the random variable that
is Sn=( X + X +…+ X)n׿ ¿, Sn is a random walk of length n. We will also denote for
k ≤ n S k as a subwalk of Sn .also S will be a random walk of length infinity (a limit of
Sn) and S(k) is a sub walk of S of length k.

Now we will define 2 sequences an and bn.

a n=P ( S n=0 ) b n=P(Sn =0 ∧ ∀0 <k<n S k ≠ 0)

We will define a0 = 1, b0 = 0.

Let A(x),B(x) be the generating functions of an and bn respectively.

Given a random walk that satisfies the rule of an it returns for some k ≤ n to the
origin for the first time and then it goes on a "random walk" of length n-k that
n
returns to the origin again. So, from that ∀n >0 an=∑ bk ∗an−k .
k=0

Also for n=0 :a0 =b0∗a1 +1, so we get A ( x )=1+ A ( x )∗B ( x )

Notice that B (1 )=∑ b n=lim P ( ∃k ≤ n Sk =0 ) =P ( ∃k S( k )=0 ) or in other words the


n →∞

probability that a random walk will return to the origin.

Notice as well, that A ( 1 )=∑ an=lim E [ ¿ k≤ n S k =0 ]=E[¿k S (k )=0] or in other words,


n→∞

the expected value for the number of times the random walk returns to the
origin.
1
As we know A(x) = A(x)*B(x) + 1 we get B ( x ) =1− by taking the limit as x
A (x)
approaches 1 we get the P(return) = 1 ⇔ E [#return] = ∞
We will now prove that for dimension = 1 E [ ¿ return ] =∞

For that we will calculate an.

Obviously for odd n an = 0 as the walk will end at an odd number.

For n = 2m the number of walks ending at 0 is exactly the number of lists with 1

and -1 such that the number of 1s is equal to the number of -1s. this is
2m
m
and ( )
the number of total walks is 22m = 4m.

= m ) we've already seen a


( 2m
So a2m 2m+1 = 0 and hence
m
4

( )
2n

E [ ¿ returns ] = A ( 1 )=∑ an=∑ a2 n=∑


( ) =∑
2m
m 2 m!
( by stirling approximation) ∑
√ 4 πn∗
2n
e

2 πn∗( ) ∗4
m
4 ( m ! )2∗4 m n
2n
n
e

1 1 1 1 1
¿∑ = ∑ ≥ ∑ n =∞ so, we got E [ ¿ returns ] =∞ , and hence
√ πn √ π √n √ π
P(return) = 1.

Now we will look at random walks on 2 dimensions. Notice we can look at the

( )( )
1 1
2 , 2
walk in the basis in this basis X is a random variable with 2
1 −1
2 2

()
1
2 ∧independently
independent choices: ± 1 for the coefficient of
1
2

()
1
± 1 for the coefficientof 2 . So we get that this is just 2 random walks of
−1
2
dimension 1. From this we get that if we denote cn as the an from the first part of
this essay for dimension 2 and an as the an for dimension 1, cn = an2.
1 1 1
E [ ¿ returns ] =∑ c n=∑ c 2n =∑ ( a2 n ) ( as we showed before for a n ) ∑ = ∑ =∞
2
π∗n π n
so as we showed P(return) = 1 for dimension 2 as well.

To prove P ( return ) ≠ 1 for dimension 3 and above we need to use way more
advanced methods (such as moment) as the trick for dimension 2 will not work.

You might also like