Scribd
Upload
57 views25 pages

Brawnian Motion

Brownian motion refers to the random movement of small particles suspended in fluids over time. It can be modeled as a continuous stochastic process where the particle's position is a random variable that changes incrementally through time via independent, normally distributed displacements. Brownian motion has the properties of starting at the origin, having independent and stationary increments, and being a Markov process. It can be used to model various random phenomena like stock price movements.

Uploaded by

DANISH ELCHI
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
57 views25 pages

Brawnian Motion

Brownian motion refers to the random movement of small particles suspended in fluids over time. It can be modeled as a continuous stochastic process where the particle's position is a random variable that changes incrementally through time via independent, normally distributed displacements. Brownian motion has the properties of starting at the origin, having independent and stationary increments, and being a Markov process. It can be used to model various random phenomena like stock price movements.

Uploaded by

DANISH ELCHI
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 25

BRAWNIAN MOTION

• a number of particles in Brownian motion are present in a given


medium and there is no preferred direction for the random movment,
then over a period of time the particles will tend to be spread evenly
throughout the medium.
• Brownian motion refers to the random movement displayed by small
particles that are suspended in fluids. 
• It is commonly referred to as Brownian movement”. This motion is a
result of the collisions of the particles with other fast-moving particles
in the
• Two-dimensional Brownian motion can be used, for example, to
model spatial movement over time of insects on a plot of land
•  one-dimensional Brownian motion has been used in finance to track
stock movements. 
• The tracking of stock movements over time is an example of an
approximation by a continuous process of an essentially discrete
process observed over a long period at small intervals.
Discrete​ Continuous​
​

R.W​ ​
Branching​

Poisson​ Brownian motion​


Queue​
B/D​
• X(t) : a continuous random variable that depends on time t 
• X(0) = 0 such that for any time interval (ti , ti+1), the increments or
changes

•  D(ti , ti+1) = [X(ti+1) − X(ti)]

• Independent, identically distributed except the dependence on (ti+1-


ti)
• 0<= i<=n, ti < ti+1

• Where n is finite: this is a Markov property. Furthermore it is assumed


that the D(ti , ti+1) have the same probability distribution for 0 ≤ i ≤ n
which depend only on the length o
• To simplify matters it is assumed that this is a process with the
expected value
•  E[X(t)] = 0: this defines a process with no drift. 

• The random variable 


• X(t) = D(0, t) 
• so that X(s) will have the same probability distribution as X(s)=   D(t,
t + s) for all s.
• Initial condition 
• X(0)=0
• D(ti , ti+1) = [X(ti+1) − X(ti)]
• X(t) = D(0, t) = X(t) - X(0)= X(t)
• X(t) can be shown to have a normal distribution.
• Generally,  a stochastic process with stationary,
independent increments is called a Levy process
The name Wiener process  is also now more common when discussing
technical det we use both terms here.
• To summarize, a Wiener process or Brownian motion for a
continuous random variable X(t) has the following properties: 

• (a) X(0) = 0; 


• (b) D(ti , ti+1) = [X(ti+1) − X(ti)]  has a normal distribution with
•  mean 0 and variance σ2 (ti+1 − ti), 1 ≤ i ≤ n − 1;
•  (c)D(ti , ti+1) are mutually independent in non-overlapping time
intervals.
• By choosing t1 = 0 and t2 = 1, then σ 2 = V[X(1)]. The process where
σ 2 = 1 is known as 

• standard Brownian motion or Wiener process. 


• Any (non-standard) process can be converted to standard process W(t)
by defining W(t) = X(t)/σ. 
• The dispersion of standard Brownian motion can be shown by viewing
computed outputs against the variance of the normal distribution. The
probability density function of standard Brownian motion is given by

• ft(x) = 1 √ 2πt exp (−x2/ 2t ) , (−∞ < x < ∞), (t > 0), .
Wiener process as a limit of a random walk 
• The symmetric one-dimensional random walk was 
• with step length ±1 . 
• The probability of a positive or negative step is 1 /2 , 
• The position of the walk, Xn, at stage n 
• has a distribution with E(Xn) = 0 and V(Xn) = n. 
• We adapt the symmetric random walk in the following way to create a
limiting process on the walk.
• Divide the time-scale t ∈ [0, ∞) into intervals of fixed (small) length
ε. At each step t = 0, ε, 2ε, . . . 

• Step is not +1 or –1 


• there is a probability 1/ 2 that the walk advances √ ε or 
• a probability 1/ 2 that the walk retreats − √ ε.
• If Xi is the random variable of the position of the walk at step i, the

• Xi = X0 + Σ Qj = Σi Qj,

• assuming X(0) = 0. 
• Now E(Qn) = 1/ 2 √ ε − 1/ 2 √ ε = 0, 
• V(Qn) = E(Q2n) − E(Qn)2 = ε,
• for all n
•  E(Xn) = 0, 
• V(Xn)  = nε,
Scaling
• Suppose that the random variable 
• X(t) : a one-dimensional Brownian motion
• X(0) = 0 . 
•  the standard Brownian motion W(t) = X(t)/σ 
• Consider now the transformation of X(t) to S(t) given by 
• S(t) = aX(t/b),
• where a ≠ 0  and b > 0 are constants.
•  S(t) is a linear function of a continuous normally distributed random
variable X(t), and so is also normally distributed but with
•  E(s(t)) = 0 and
• Var(s(t)=a2σ2 t/b
• A particular case of this transformation occurs when
•  a = −1 and b = 1.
•  This is known as the reflection of the original process, with the same
properties as the original.
• We can considering the particular transformation
•  a = √ c and b = c, 
• where c is a positive constant, 
• giving rise to the Brownian motion Q(t) = √ cX(t/c)
• . In other words the distributional properties of Q(t) are the same as
those of X(t) whatever the value of the constant c. 
• It does not matter how much the two axes are stretched given increasing
detail of the path: Q(t) still acts in a similar way to the original path
• . This parallels a phenomenon known as a fractal.
First visit times

You might also like