Brownian motion

Girish S. Setlur

1. Introduction
Here is a detailed explanation of Discrete Brownian Motion (also known as the random walk) in 1D and 2D, including definitions,
formulas, and fully worked-out examples.

1. Discrete brownian motion (random walk): overview

A discrete random walk is a simplified model of Brownian motion where a particle takes steps at discrete time intervals, each of
fixed length, in random directions. It is a basic model used to study diffusion, Brownian motion, and statistical physics.

2. 1-dimensional discrete random walk

Definition:
A particle starts at position x = 0 at time t = 0. At each time step, it moves:
One step right with probability p One step left with probability q = 1 − p
Let:
Xn be the position after n steps. For symmetric walk, p = q = 12 .
Then:
n
X
Xn = ξi
i=1

Where:
ξi = +1 with probability p, and −1 with probability q

Properties:
Mean position:
n
X
hXn i = < ξi >= n(p − q)
i=1

Variance:
n
X
hXn2 i − hXn i2 = (< ξi ξj > − < ξi >< ξj >)
i,j=1
n
X n
X
= (< ξi ξi > − < ξi >< ξi >) = (1− < ξi >2 )
i=1 i=1
2
= n(1 − (p − q) ) = n(1 − p + q)(1 + p − q) = 4npq

For symmetric walk (p = q = 21 ):

hXn i = 0, Var(Xn ) = n

Probability distribution (binomial):

 
n n+k n−k
P (Xn = k) = n+k p 2 q 2 , where n + k even
2

1
Example 1: Symmetric 1D Random Walk
Let a particle take 4 steps with equal probability in each direction (p = q = 21 ).
List all possible paths and final positions:

Path Steps Final Position

RRRR +1+1+1+1 +4
RRRL +1+1+1-1 +2
RRLR +1+1-1+1 +2
RLRR +1-1+1+1 +2
LRLR et.c ... ...
4
Total number of paths: 2 = 16
Probability of ending at position 0:
To reach position 0 in 4 steps:
Must have 2 right and 2 left steps.
   4
4 1 1 3
P (X4 = 0) = =6· =
2 2 16 8

Example 2: Asymmetric 1D Walk

Let p = 0.6, q = 0.4, and n = 5. What is the probability the walker ends at X5 = +1?
Let r be the number of right steps, and l = 5 − r the number of left steps.
Then X5 = r − l = 2r − 5 = +1 ⇒ r = 3
 
5
P (X5 = +1) = (0.6)3 (0.4)2 = 10 · 0.216 · 0.16 = 0.3456
3

3. 2-dimensional discrete random walk

Definition:
A particle starts at the origin (x, y) = (0, 0). At each step, it moves:
Up (0, +1) Down (0, −1) Right (+1, 0) Left (−1, 0)
Each with probability 14 (symmetric walk on square lattice).
Let:

~ n = (Xn , Yn )
X

Properties:
Expected position:

hXn i = hYn i = 0

Mean square displacement:

~ n |2 i = hX 2 + Y 2 i = n · hstep length2 i = n · 1 = n
h|X n n

Probability distribution: Multinomial (can be approximated by 2D Gaussian for large n).

Example 3: 2D Random Walk Probability of Returning to Origin in 2 Steps

How many paths return to origin in 2 steps?
Possible 2-step paths:

2
Up Down Down Up Left Right Right Left
Each of these 4 combinations brings the particle back to (0,0). Total number of paths = 42 = 16
Number of paths returning to origin = 4

~ 2 = (0, 0)) = 4 = 1
P (X
16 4

Example 4: Distribution After 4 Steps in 2D

How many paths lead to position (0, 0) after 4 steps?
To return to origin in 4 steps:
Must have equal number of left-right and up-down steps.
Let:
nR = nL = r, nU = nD = u, with r + u = 2
So:
4!
Number of such paths = , for r + u = 2
r!r!u!u!

Only combinations:
r = 2, u = 0: 4!/(2!2!0!0!) = 6 r = 1, u = 1: 4!/(1!1!1!1!) = 24 r = 0, u = 2: 4!/(0!0!2!2!) = 6
Total = 6 + 24 + 6 = 36
Total paths = 44 = 256

~ 4 = (0, 0)) = 36 = 9
P (X
256 64

4. Connection to continuous Brownian motion

Discrete random walk becomes continuous Brownian motion in the limit:

(δx)2
δx → 0, δt → 0, with = 2D constant
δt

The probability distribution converges to the Gaussian:

x2
 
1
P (x, t) = √ exp −
4πDt 4Dt

5. Summary table
Feature 1D RW 2D RW
————————- —————————– ——————————-
Possible directions Left, Right Up, Down, Left, Right
Position after n steps Xn =
P
ξi ~ n = P ξ~i
X
Mean hXn i = 0 hX~ ni = 0
Variance / MSD hXn2 i = n ~
h|Xn |2 i = n
Distribution Binomial ( Gaussian) Multinomial ( 2D Gaussian)
Return to origin (probability decreases with n) Yes Yes (but slower decay)

5. Derivation of Gaussian distribution as a limiting case of binomial distribution

Consider a 1D symmetric random walk. At each time step:
1
• Move right by +δx with probability p = 2
1
• Move left by −δx with probability q = 2

3
 After n steps, the total displacement is
n
X
Xn = ξi , where ξi = ±δx.
i=1
Equivalently,
Xn = (2r − n)δx,
where r is the number of right steps.

The probability of r right steps is given by the binomial distribution:

   n
n 1
Pn (r) = .
r 2
We relate r to the position x as
x n
x = Xn = (2r − n)δx ⇒ r = + .
2δx 2
So,    n
n 1
P (x, n) = n x .
2 + 2δx 2

Using Stirling’s approximation for large n,

1
ln(n!) ≈ n ln n − n + ln(2πn),
2
we approximate the binomial coefficient and find
(r − np)2
 
1
P (x, n) ≈ √ exp − .
2πnpq 2npq
In the symmetric case (p = q = 12 ), this becomes
x2
 
1
P (x, n) ≈ √ exp − .
πn 2n(δx)2
t
Let t = nδt, so n = δt . Then
x2
 
1
P (x, t) ≈ p exp − .
2π(δx)2 t/δt 2(δx)2 t/δt

Define the diffusion coefficient D as

(δx)2
D = lim .
δt→0 2δt
So,
x2
 
1
P (x, t) ≈ √ exp − .
4πDt 4Dt

The limiting distribution P (x, t) satisfies the diffusion equation:

∂P ∂2P
=D 2,
∂t ∂x
with initial condition P (x, 0) = δ(x).

Discrete Random Walk Continuous Brownian Motion

Steps ±δx Continuous paths
Time step δt Continuous time
Binomial distribution Gaussian distribution
Variance n(δx)2 Variance 2Dt
Diffusion limit Satisfies diffusion equation