Your Paper

You
December 7, 2023

Abstract
Your abstract.

1 Introduction
The image you’ve uploaded contains handwritten notes on stochastic integrals, specifically those related
to the Wiener process, which is a mathematical model for Brownian motion. Here’s a transcription of
the notes:
—
**Itō stochastic integrals**
Z t n
X
′ ′
δ(t )dW (t ) = lim δ(t∗i )[W (ti ) − W (ti−1 )]
t0 n→∞
i=1

This represents the Itō integral of the delta function against the Wiener process, which is a mathe-
matical model of Brownian motion.
**Wiener process** The Wiener process, denoted W (t), is a continuous-time stochastic process
with stationary, independent increments and is often used to model random paths such as stock prices
or physical Brownian motion.
Z t
W (t′ )dW (t′ ) =?
t0

This is asking for the solution to the Itō integral of the Wiener process with respect to itself.
Z t
1 2  1
W (t′ )dW (t′ ) = W (t) − W 2 (t0 ) − (t − t0 )
t0 2 2
This is the solution to the above integral, which is a well-known result in stochastic calculus. The first
term is the net change in the square of the Wiener process, and the second term corrects for the Itō
drift.
Z t 2 Z t
′ ′
⟨ W (t )dW (t ) ⟩ = ⟨W 2 (t′ )⟩dt′
t0 t0

1 2  1
= W (t) − W 2 (t0 ) − (t − t0 )
2 2
Here, the calculation considers the square of the integral, which involves taking the expected value
over the square of the Wiener process increments.
Z t
1 1
⟨W (t′ )dW (t′ )⟩ = ⟨W 2 (t) − W 2 (t0 )⟩ − (t − t0 )
t0 2 2
This shows the expected value of the Itō integral itself, which involves the expectation of the difference
in the squares of the Wiener process at two different times. Certainly, here is the transcription of the
notes from the image:
—

1
 1) Itō integration is different from Riemann integration. The rules are different.
**Choice 2** ζ = 12
Z t
W (t′ )dW (t′ )
t0
n  
X 1
= lim [W (ti ) + W (ti−1 )] (W (ti ) − W (ti−1 ))
n→∞
i=1
2
n
1X 2
= ⟨ms-limit (W (ti ) − W 2 (ti−1 ))⟩
2 i=1
1 2
= [W (t) − W 2 (t0 )]
2
(Stratonovich)
—
Here is the transcription of the notes from the image you’ve uploaded:
—
By choosing α we change the integrator 1) α = 0 τ̂i = ti−1 ... Itō stochastic integral
2) α = 21 τ̂i = ti +t2i−1 ... Stratonovich integral
3) α = 1 Klimontovich (Hänggi)

τ̂i = αti + (1 − α)ti−1

—
The notes discuss how the choice of α in the definition of τ̂i changes the interpretation of the
stochastic integral. The three cases mentioned correspond to different types of stochastic calculus:
1) With α = 0, τ̂i is chosen as ti−1 , which corresponds to the Itō stochastic integral. 2) With
α = 21 , τ̂i is the midpoint between ti and ti−1 , leading to the Stratonovich integral. 3) With α = 1,
the notes mention Klimontovich, also known as the Hänggi integral, which is another interpretation
used in stochastic processes.
These choices affect how the integral of a stochastic process is computed and interpreted, with Itō
being non-anticipative and Stratonovich following the more classical chain rule from calculus. The
Klimontovich or Hänggi integral is less commonly used but is important in certain areas of physical
sciences and engineering.
Certainly, here is the transcription of the notes from the image:
—
**Itō Integration:**
Itō integration is a key concept in stochastic calculus, particularly when dealing with Brownian
motion represented by a Wiener process W (t). Unlike classical Riemann integration, Itō integration
accounts for the stochastic nature of the integrand.
More precisely, for the square of a differential of W (t) we have:

(dW (t))2 = dt

and for powers higher than two, it holds that:

(dW (t))n = 0 for n > 2

For the integral of the square of the differential of W (t) multiplied by a function G(t), we consider
the mean-square limit as follows:
Z t X
(dW (t′ ))2+N G(t′ ) = lim Gi−1 ∆Wi2
t0 n→∞
i
Z t
= dt′ G(t′ ) for N = 0
t0

and this limit equals zero for N > 0.

2
 When N = 0, the integral I is defined as:
* +
X
2


I = lim Gi−1 (∆Wi ) − ∆ti
n→∞
i
* +
X
G2i−1 2


= lim (∆Wi ) − ∆ti
n→∞
i

where Gi−1 is the value of the function G(t) at the previous time step, ti−1 , and ∆Wi are the increments
of the Wiener process between ti−1 and ti .
The integral I can then be expressed as:
X
I = 2 lim ∆ti G2i−1
n→∞
i

which utilizes the property that ⟨(∆Wi ) ⟩ = ∆ti and ⟨(∆Wi )2 −∆ti ⟩2 = 2∆t2i for the Gaussian process
2

W (t), where ∆Wi is independent of ∆Wj for i ̸= j.

— Here is the transcription of the notes from the image provided:
—
**Itō Integration:**
The mean-square (ms) limit of the sum of products of Gi−1 and the square of Wiener process
increments ∆Wi minus the time increments ∆ti approaches zero:
X X
ms-limit Gi−1 ∆Wi2 − Gi−1 ∆ti = 0
i i

Similarly, the mean-square limit of the sum of Gi−1 times ∆ti is equal to the Itō integral of G(t′ )
times the square of the Wiener process increment:
X Z t Z t
ms-limit Gi−1 ∆ti = dt′ G(t′ ) = (dW (t′ ))2 G(t′ )
i t0 t0

**Properties of Itō Stochastic Integrals:** Rt

1. **Existence**: The Itō stochastic integral t0 G(t′ )dW (t′ ) exists if G(t) is a continuous and
non-anticipating function.
A non-anticipating function G(t) means it is statistically independent of future increments of the
Wiener process W (s) − W (t) for s ̸= t.
— Here’s the transcription and clarification of the notes from the image:
—
**2)** The differential of a function of W (t) raised to the n-th power is given by:
n
d(W̃ (t))n = [W (t) + dW (t)] − W (t)n

n  
X n
= W (t)n−r (dW (t))r
r=1
r
n(n − 1)
= nW (t)n−1 dW (t) + W (t)n−2 (dW (t))2 + higher-order terms
2
**3)** For an exponential function of W (t), the differential is:
**Itō’s Lemma:** For a general function f (W (t), t), Itō’s lemma provides the differential in terms
of partial derivatives:

∂f ∂f 1 ∂2f
df (W (t), t) = dt + dW (t) + (dW (t))2
∂t ∂W 2 ∂W 2
This can be further expressed as:

1 ∂2f
 
∂f ∂f
df (W (t), t) = + dt + dW (t)
∂t 2 ∂W 2 ∂W

3
Apologies for the confusion. Based on your correction, the stochastic differential of the exponential of
a Wiener process W (t), without any multiplicative constant ρ, is:
 
dt
d exp(W (t)) = exp(W (t)) dW (t) +
2
This formula is derived using Itō’s lemma, which in its general form for a function f (W (t), t) is:

∂f ∂f 1 ∂2f
df (W (t), t) = dt + dW (t) + (dW (t))2
∂t ∂W 2 ∂W 2
Applying this to f (W (t)) = exp(W (t)):
1. The first derivative of f with respect to W is exp(W (t)). 2. The second derivative of f with
respect to W is also exp(W (t)). 3. The derivative of f with respect to t is 0 since there is no explicit
time dependence.
Putting these into Itō’s lemma:
1
d exp(W (t)) = exp(W (t))dW (t) + exp(W (t))(dW (t))2
2
Since (dW (t))2 = dt, the equation becomes:
1
d exp(W (t)) = exp(W (t))[dW (t) + exp(W (t))dt
2
Which matches the expression you’ve provided. **4)** For non-anticipating functions G(t′ ), the
expected value of the integral from t0 to t of G(t′ ) times dW (t′ ) is zero:
Z t 
G(t′ )dW (t′ ) = 0
t0

—
These notes discuss how to apply Itō’s calculus to different functions of Brownian motion, specif-
ically focusing on the Itō’s lemma which is a key result in stochastic calculus that allows one to
differentiate and integrate functions of stochastic processes. Itō’s lemma relates the stochastic differ-
ential of a function to its partial derivatives, enabling the calculation of expectations for stochastic
integrals involving non-anticipating functions. The image contains a handwritten note that states an
important property of stochastic integrals in the context of Itō calculus. Here is the transcribed content
of the note:
—
**5)** If G(t) and H(t) are arbitrary continuous non-anticipating functions, then the expected
value of the product of their Itō integrals with respect to a Wiener process W (t) from t0 to t is given
by the Lebesgue integral of the product of the functions’ expected values:
Z t Z t  Z t
′ ′ ′ ′
G(t )dW (t ) H(t )dW (t ) = ⟨G(t′ )H(t′ )⟩ dt′
t0 t0 t0
—
This expresses a crucial concept in stochastic calculus: the covariance of the Itō integrals of two
non-anticipating functions over the same interval with respect to the same Wiener process is equal
to the integral of their covariance. Non-anticipating functions ensure that the value of the function
at a certain time does not depend on the future values of the Wiener process. This allows for the
interchange of the expectation and integration when dealing with the product of Itō integrals. The
content of the slide is as follows:
—
**Stochastic differential equations**
Alternatively, in probability theory finance the SDEs are written as

dX(t) = µ(X(t), t)dt + σ(X(t), t)dW (t)

cf. y ẏ = Fdet (t) + Frand (t) (Measure of a random process, for instance, Wiener measure)

4
 Formally, a process X(t) satisfies a stochastic differential equation

dX(t) = µ(X(t), t)dt + σ(X(t), t)dW (t), t ∈ [0, T ]

iff it satisfies the following stochastic integral equation

Z t Z t
X(t) = X(0) + µ(X(s), s)ds + σ(X(s), s)dW (s)
0 0

t ∈ [0, T ]
—
This slide presents the standard form of a stochastic differential equation (SDE) and its associated
integral form, which is used to describe the evolution of a stochastic process over time. Here is the
transcription of the text from the image you’ve uploaded:
—
**Stochastic differential equations**
One can conveniently use the SDEs to simulate the trajectory of a process

ẏ = A(y) + ξ(t)
⟨ξ(t)⟩ = 0
⟨ξ(t )ξ(t)⟩ = qδ(t − t′ )
′

p
y(t + dt) = y(t) + A(y)dt + 2qdtZ
Z is random variable drawn from N(0,1)

ẋi = fi (x) + gi (x)ξ(t), i = 1, N

⟨ξ(t)⟩ = 0, ⟨ξ(t′ )ξ(s)⟩ = δ(t − s)

√ √
x(h) − x(0) = 2DZ1 + f h + f ′ v 2DZ2
1
+ f f ′ h2 + f DZ3 f ′′
2
Fast and precise algorithm for computer simulation of stochastic differential equations, R. Mannella,
V. Palleschi, Phys Rev A, 3381 (1989)
The image appears to be a slide from a presentation on the Ornstein-Uhlenbeck process, which is
a type of stochastic differential equation. Here’s a transcription of the content on the slide:
—
**Skoltech** Stochastic differential equations
Ornstein-Uhlenbeck process

dX = −γXdt + σdW
Diffusion in a harmonic potential
Z t
X(t) = e−γt X(0) + σ e−γ(t−s) dW,
0
—
The slide features the differential equation for the Ornstein-Uhlenbeck process, which is a model for
the velocity of a particle undergoing Brownian motion in a fluid with friction. The process describes
the evolution of a system experiencing a restoring force towards a mean value, proportional to the
deviation from that mean (described by the term −γX), and a stochastic force represented by σdW ,
where dW is the increment of a Wiener process.
The solution to the Ornstein-Uhlenbeck process is given, showing how the value of the process at
time t, X(t), depends on its initial value X(0) and a stochastic integral representing the accumulated
effect of the random shocks over time.

5
 The slide also includes a graphical representation of an Ornstein-Uhlenbeck process and a potential
well, illustrating the concept of diffusion in a harmonic potential. This is a common way to visual-
ize the Ornstein-Uhlenbeck process, as it is akin to a particle moving in a quadratic potential well
under the influence of random noise. The image provided appears to be a slide from a presentation
discussing stochastic differential equations (SDEs) in the context of financial modeling, specifically
the Black-Scholes model for option pricing, which uses Geometric Brownian Motion (GBM). Here’s a
transcription of the content:
—
**Skoltech** Stochastic differential equations
Geometric Brownian motion example (Black–Scholes model for option pricing)
dS
= µ(S(t), t)dt + σ(S(t), t)dW (t)
S
Samuelson (1965) and Merton (1973)
(price) (return on the stock)
Itō calculus solution
Z t  Z t
σ 2 (s)
 
S(t) = S(0) exp µ(s) − ds + σ(s)dW (s)
0 2 0

if µ(s) = const, σ(s) = const

σ2
 
S(t) = S(0) exp (µ − )t + σW (t)
2
—
The slide is explaining the SDE that models the price of a stock using GBM, where S is the stock
price, µ is the expected return, σ is the volatility, and W (t) is a Wiener process or Brownian motion.
The differential equation dS/S represents the proportional or percentage change in stock price, which
is modeled by a drift component µdt and a diffusion component σdW (t).
The solution to this SDE is given in terms of the initial stock price S(0) and an exponential
function incorporating both the drift and diffusion terms. If the drift µ and volatility σ are constants,
the solution simplifies to the second exponential formula, which reflects the log-normal distribution of
stock prices according to the Black-Scholes model.
This model is foundational in financial mathematics and is used for pricing options, a type of
financial derivative. It was developed by Fisher Black, Myron Scholes, and Robert Merton in the early
1970s. The work of Paul Samuelson also contributed to the development of the model. The image
seems to be a slide from a presentation on probabilistic modeling and provides an alternative approach
to a mathematical problem. Here’s a transcription of the content from the image:
—
**Skoltech - What is an alternative?**
Z
P (x3 , t3 |x1 , t1 ) = dx2 P (x3 , t3 |x2 , t2 )P (x2 , t2 |x1 , t1 )

(1) P (x, t + τ |0, 0) = P (x, t + τ ) = dx′ P (x, t + τ |x′ , t)P (x′ , t)

R

Introducing ∆ = x − x′ and performing Taylor expansion

(2) P (x, t + dτ |x′ , t)P (x′ , t) = P (x − ∆ + ∆, t + dτ |x − ∆, t)P (x − ∆, t)
∞  n
X (−1)n n ∂
= ∆ P (x + ∆, t + dτ |x, t)P (x, t)
n=0
n! ∂x

Inserting (2) into (1) and integrating we get (see the next slide):
—
This slide appears to be discussing a method for dealing with transition probabilities in a stochastic
process. It begins with the Chapman-Kolmogorov equation, which relates the transition probabilities
of states over time. The slide then introduces the concept of the Taylor expansion in the context of
these probabilities and sets up an equation that will be further elaborated upon in the next slide of
the presentation.

6
 The Chapman-Kolmogorov equation is a foundational concept in the theory of stochastic processes
and is used to calculate the probability of transitioning from one state to another over a certain period
of time, given the transition probabilities for intermediate times. The Taylor expansion is used to
approximate the change in probability over a small time increment dτ , which is crucial for deriving
differential equations that describe the evolution of probability distributions, such as the Fokker-Planck
equation or the Master equation. The image you’ve provided is a slide from a presentation detailing the
Kramers-Moyal expansion, which is a mathematical technique used in the field of stochastic processes.
Here’s the transcription of the slide content:
—
**Skoltech Kramers-Moyal expansion**
∞  n
X ∂ Mn (x, t, τ )
P (x, t + τ ) − P (x, t) = − P (x, t),
n=1
∂x n!
where
Z
Mn (x, t, τ ) = (x − x′ )n P (x, t + τ |x′ , t)dx′

Now, we expand the moments Mn into Taylor with respect to τ

Mn (x, t, τ )/n! = D(n) (x, t)τ + O(τ 2 )

Finally, we arrive at the equation for the pdf!
∞  n
∂P (x, t) X ∂
= − D(n) (x, t)P (x, t) = LKM P (x, t)
∂t n=1
∂x
—
This slide presents the Kramers-Moyal expansion, which is an infinite series that expresses the time
evolution of the probability density function P (x, t) of a stochastic process. The Mn (x, t, τ ) terms rep-
resent the moments of the transition probabilities, and the D(n) (x, t) are the so-called Kramers-Moyal
coefficients. The equation highlighted at the bottom is the partial differential equation resulting from
truncating the Kramers-Moyal expansion, which is used to describe the probability density function
(pdf). The operator LKM is the Kramers-Moyal operator which, when applied to P (x, t), gives the
time derivative of the probability density function. This expansion is closely related to the Fokker-
Planck equation when truncated to the second term. The image you’ve uploaded is a slide from a
presentation explaining the Fokker-Planck equation, which is a partial differential equation describing
the time evolution of the probability density function of the velocity of a particle under the influence of
drag forces and random forces, commonly used in statistical mechanics, and other fields dealing with
stochastic processes. Here’s the content of the slide:
—
**Skoltech The Fokker-Planck equation**
If we stop the Kramers-Moyal expansion after 2 terms we get

∂ 2 (2)
 
∂P (x, t) ∂ (1)
= − D (x, t) + D (x, t) P (x, t) = LF P P (x, t)
∂t ∂x ∂x2
∂P (x, t) ∂S(x, t)
+ = 0,
∂t ∂x
PDF ← Probability current ↓
 
∂ (2)
S(x, t) = D(1) (x, t) − D (x, t) P (x, t)
∂x
—
The slide shows the Fokker-Planck equation in one dimension, where P (x, t) is the probability
density function for the position x of a particle at time t, D(1) (x, t) is the first drift coefficient, and
D(2) (x, t) is the diffusion coefficient. The operator LF P is the Fokker-Planck operator. This equation
is derived by stopping at the second term of the Kramers-Moyal expansion.

7
 The second equation on the slide expresses the conservation of probability, where S(x, t) is the
probability current, which must satisfy the continuity equation, implying that the change in probability
density over time plus the divergence of the probability current is zero.
The slide suggests that the Fokker-Planck equation is an alternative approach to describing systems
that would otherwise be modeled using stochastic differential equations (SDEs). The Fokker-Planck
equation provides a more detailed and rigorous way of looking at the time evolution of probability
distributions, offering insights into the diffusive processes that cannot be obtained from SDEs alone.
The content of the image describes the Fokker-Planck equation and its application to the Wiener
process, which is a fundamental concept in stochastic processes and is extensively used in physics and
finance. Here’s the transcription of the content:
—
**Fokker-Planck Equation:**
The general form of the Fokker-Planck equation is given by:

∂ 2 (2)
 
∂P (x, t) ∂ (1)
= − D (x, t) + D (x, t) P (x, t)
∂t ∂x ∂x2

**Examples:**
Wiener process
∂P (x, t) ∂ 2 P (x, t)
=D
∂t ∂x2
with the initial condition
P (x, t|x′ , t′ ) = δ(x − x′ )
The solution to this equation with the given initial condition is:

x2
 
1
P (x, t) = √ exp −
4πDt 4Dt
—
The Fokker-Planck equation is used to describe the time evolution of the probability density func-
tion P (x, t) of the position x of a particle at time t. The drift coefficient D(1) (x, t) and the diffusion
coefficient D(2) (x, t) determine the behavior of the process.
For the Wiener process, which is a simple stochastic process with no drift and constant diffusion, the
Fokker-Planck equation reduces to the diffusion equation shown. The solution provided is a Gaussian
function which is the fundamental solution to the diffusion equation, representing the probability
density function of the particle’s position at time t given that it started at position x′ at an earlier
time t′ . The δ function indicates that the particle was initially at x′ with certainty (i.e., the initial
condition is sharply peaked at x′ ). The parameter D is the diffusion coefficient, which quantifies the
rate at which the probability spreads out over time.