Scribd
Upload
241 views19 pages

Simple Symmetric Random Walk: Reference: Feller, Volume I, Chapter 3

1. A simple symmetric random walk is a stochastic process where the next step is randomly either +1 or -1 with equal probability, starting from 0. This can model situations like a fair gambling game. 2. While the random walk is certain to return to the origin infinitely often, the expected time between returns is infinite. This is an example of null recurrence, which can cause paradoxes like both people in a store having infinite expected wait times. 3. The frequency of the first return to the origin, T, can be calculated using the key lemma that the probability of staying away from the origin for an even number of steps is equal to the probability of being at the origin after that many steps. However

Uploaded by

Rishi Mohan
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
241 views19 pages

Simple Symmetric Random Walk: Reference: Feller, Volume I, Chapter 3

1. A simple symmetric random walk is a stochastic process where the next step is randomly either +1 or -1 with equal probability, starting from 0. This can model situations like a fair gambling game. 2. While the random walk is certain to return to the origin infinitely often, the expected time between returns is infinite. This is an example of null recurrence, which can cause paradoxes like both people in a store having infinite expected wait times. 3. The frequency of the first return to the origin, T, can be calculated using the key lemma that the probability of staying away from the origin for an even number of steps is equal to the probability of being at the origin after that many steps. However

Uploaded by

Rishi Mohan
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 19

Simple Symmetric Random Walk

Reference: Feller, Volume I, Chapter 3.


We will illustrate some important methods,
using the sample path (geometric) approach.
We will illustrate some of the important
concepts of stochastic processes.
Definition and Notation
n n
i i
X X S
S
i X P X P
X X X
+ + =
=
= = = = =

1
0
3 2 1
0
, 3 , 2 , 1 all for 2 / 1 ] 1 [ ] 1 [
d distribute - y identicall and t independen are , , ,
All sample paths are equally likely.
Plot S
n
as
a function
of n.
Essentially binomial distribution at a fixed time, but translated.
Applications
A sequence of fair gambling games.
Comparing two waiting lines.
The price of a stock on the NYSE.
The random path of a particle being
bombarded by other particles.
Possibly Interesting Questions
How often does the path return to the origin?
What is the frequency function of the first return?
What is the expected time between returns?
What proportion of the time does the path
spend positive?
What proportion of time are you winning at a fair
game?
How long does it take until the first time the
path hits level k?
Probability of Return to Origin
origin) the of time" hitting " (or
origin the return to first of time
never) if ( } 0 : 0 min{
n
1

2
2
] 0 [
2
2
=
= > =
~
|
|
.
|

\
|
= =

n
n
n
S n T
n
n
S P
t
By Stirlings formula.
This requires an even time.
Two Important Sequences
1
2
2
] 0 [
] 2 [
0
2
2 2
2
=
|
|
.
|

\
|
= = =
= =

u
n
n
S P u
n T P f
n
n n
n
By convention.
We start at zero.
Unknown frequency function of T.
Periodicity
Recursion Approach
l n
n
l
l n
u f u
2 2
1
2 2
=

=
This recursion formula could be used to solve inductively for
the values of the frequency function f, but we would not
necessarily get a explicit functional form for the values in
general.
We will prove this, but not proceed further using this approach.
First Entrance Decomposition
This is a common method in stochastic processes.
] 0 [ ] 2 [
] 0 [ ] 2 [ ] 0 [
} 0 , 0 , 0 , , 0 , 0 {
} 0 , 2 { } 0 {
1
2 2
2
1
2 2
2 2 2 2 2 4
1
2
1
2 2
= = =
= = = =
= = = = = =
= = = =

=
=
n
l
l n
l
n
l
n n
l n l l
n
l
n
l
n n
S P l T P
S S P l T P S P
S S S S S S
S l T S

Why write it like this?


Essentially, conditioning on the first return to zero.
A Geometric (Sample Path) Approach
n n
u S S S P
LEMMA KEY
2 2 2 1
] 0 , , 0 , 0 [
:
= = = =
The probability of staying away from zero for an even period of
time is exactly the same as the probability of being at zero at the
end of the period!
Intuitively, the first probability is going to zero, since as the time
period gets longer, there is higher probability that it will touch
zero sometime. We already know the second probability goes to
zero.
Step 1 of Proof
] 0 , , 0 [ 2 ] 0 , , 0 [
2 1 2 1
> > = = =
n n
S S P S S P
Why? Discrete continuity
What starts positive stays positive.
Step 2 of Proof
] 0 , , 0 [
2
1
] 0 , , 0 [
1 2 1 2 1
> > = > >
n n
S S P S S P
First step up. Others dont go below one.
Consider a new set of axes
Step 3 of Proof
] 0 , 0 , , 0 [ ] 0 , , 0 [
2 1 2 1 1 2 1
> > > = > >
n n n
S S S P S S P
8
The path must be at an even height at even times, and
an odd height at odd times. At time 2n-1, it must be at height
one or greater, so stays at or above zero at time 2n.
Combine the First Three Steps
]. 0 [ to equal is right on the quantity
that the show the if finished be will we So,
] 0 , , 0 [ ] 0 , , 0 [
2
2 1 2 1
=
> > = = =
n
n n
S P
S S P S S P
Step 4: A 1-1 Correspondence
events. two e between th
ence correspond one - to - one a show to
suffices it likely, equally are paths all Since
] 0 , , 0 [ ] 0 [
2 1 2
> > = =
n n
S S P S P
Proof by Picture
A path returning to zero at the end.
A path saying at or above zero.
The first time the minimum
value is hit.
Reflect
and
translate
Translate
to the
origin
We can reverse the process, by going backward until the first time we hit the middle value.
Calculate the Frequency of T
n
n
n n
n n
n n
n
n
n
n
n
n
n n
n
n
n
n
n
u u
S S P S S P
n T P n T P f
2
2
2
2 2 2
2 2 2
2 1 2 2 1
2
2
2
1 2
1
2
2
1
) 2 )( 1 2 (
4
2
2
2
1
2 2
] 0 , , 0 [ ] 0 , , 0 [
] 2 [ ] 2 2 [

|
|
.
|

\
|

=
|
|
.
|

\
|
|
|
.
|

\
|

=
|
|
.
|

\
|

|
|
.
|

\
|

=
=
= = = = =
> > =

By the Key Lemma
A Return to the Origin is Certain
. 1 ] [
1
1 ] 2 [
Lemma, Key By the
2
= <

= s
T P
u n T P
n
In stochastic processes, we prove that it is certain to have
infinitely many returns to the origin.
The idea is to start over from the first return, and argue that we
are in the same situation as before, so a second return is certain,
and so on, and so on. There are some technicalities, though.
This is the concept of recurrence
We return to the state infinitely
often with probability one.
Paradoxes
Null recurrence is at the root of many
paradoxes relating to random processes.
In a fair gambling game, you are certain to break
even, but the expected time until it happens is
infinite.
Suppose you and another person arrive at the two
check-out stands in a store at the same time, and
choose different lines. Each person has infinite
expected waiting time until they get ahead.
Expected Waiting Time until Return

= = =
~
|
|
.
|

\
|

= =
1
2
3
2
] 2 [ 2 ] [
2
1
2
2
1 2
1
] 2 [
n
n
n T nP T E
n
n
n
n
n T P
t
The process is certain to return to the origin, but the expected waiting time is infinite!
This is the concept of null recurrence by considering the reciprocal, the expected
percentage of time spent in the state is zero.

You might also like