Lecture 12: Diffusion Processes and Stochastic Differential Equations

Similar documents
Lecture 4: Ito s Stochastic Calculus and SDE. Seung Yeal Ha Dept of Mathematical Sciences Seoul National University

I forgot to mention last time: in the Ito formula for two standard processes, putting

The concentration of a drug in blood. Exponential decay. Different realizations. Exponential decay with noise. dc(t) dt.

Stochastic Differential Equations.

Kolmogorov Equations and Markov Processes

SDE Coefficients. March 4, 2008

MA8109 Stochastic Processes in Systems Theory Autumn 2013

Some Terminology and Concepts that We will Use, But Not Emphasize (Section 6.2)

1. Stochastic Processes and filtrations

Introduction to Diffusion Processes.

Bernardo D Auria Stochastic Processes /10. Notes. Abril 13 th, 2010

Stochastic differential equation models in biology Susanne Ditlevsen

Bernardo D Auria Stochastic Processes /12. Notes. March 29 th, 2012

MATH4210 Financial Mathematics ( ) Tutorial 7

Introduction to numerical simulations for Stochastic ODEs

Clases 11-12: Integración estocástica.

Brownian Motion. An Undergraduate Introduction to Financial Mathematics. J. Robert Buchanan. J. Robert Buchanan Brownian Motion

Partial Differential Equations with Applications to Finance Seminar 1: Proving and applying Dynkin s formula

A Concise Course on Stochastic Partial Differential Equations

Continuous Time Finance

Verona Course April Lecture 1. Review of probability

Stochastic Differential Equations. Introduction to Stochastic Models for Pollutants Dispersion, Epidemic and Finance

UNCERTAINTY FUNCTIONAL DIFFERENTIAL EQUATIONS FOR FINANCE

Question 1. The correct answers are: (a) (2) (b) (1) (c) (2) (d) (3) (e) (2) (f) (1) (g) (2) (h) (1)

Large Deviations Principles for McKean-Vlasov SDEs, Skeletons, Supports and the law of iterated logarithm

1. Stochastic Process

Stochastic Calculus for Finance II - some Solutions to Chapter VII

Stochastic Differential Equations

TMS165/MSA350 Stochastic Calculus, Lecture on Applications

The Wiener Itô Chaos Expansion

Mathematical Methods for Neurosciences. ENS - Master MVA Paris 6 - Master Maths-Bio ( )

1 Brownian Local Time

Lecture 21: Stochastic Differential Equations

DISCRETE-TIME STOCHASTIC MODELS, SDEs, AND NUMERICAL METHODS. Ed Allen. NIMBioS Tutorial: Stochastic Models With Biological Applications

Poisson Jumps in Credit Risk Modeling: a Partial Integro-differential Equation Formulation

Lecture 4: Introduction to stochastic processes and stochastic calculus

Numerical methods for solving stochastic differential equations

Stochastic Integration and Stochastic Differential Equations: a gentle introduction

A Parametric Analytical Diffusion Model for Indoor Ultra-Wideband Received Signal

Some Properties of NSFDEs

Introduction Optimality and Asset Pricing

Exercises. T 2T. e ita φ(t)dt.

Stochastic Calculus and Black-Scholes Theory MTH772P Exercises Sheet 1

A new approach for investment performance measurement. 3rd WCMF, Santa Barbara November 2009

Stochastic Integration and Continuous Time Models

Short-time expansions for close-to-the-money options under a Lévy jump model with stochastic volatility

Jump-type Levy Processes

p 1 ( Y p dp) 1/p ( X p dp) 1 1 p

Lecture on Parameter Estimation for Stochastic Differential Equations. Erik Lindström

Weak solutions of mean-field stochastic differential equations

Numerical Solutions of Stochastic Differential Equations

Properties of an infinite dimensional EDS system : the Muller s ratchet

The Pedestrian s Guide to Local Time

Backward Stochastic Differential Equations with Infinite Time Horizon

Stochastic Differential Equations

On continuous time contract theory

Stochastic Processes and Advanced Mathematical Finance

MSH7 - APPLIED PROBABILITY AND STOCHASTIC CALCULUS. Contents

Stability of Stochastic Differential Equations

Homework # , Spring Due 14 May Convergence of the empirical CDF, uniform samples

APPLICATION OF THE KALMAN-BUCY FILTER IN THE STOCHASTIC DIFFERENTIAL EQUATIONS FOR THE MODELING OF RL CIRCUIT

PROBABILITY: LIMIT THEOREMS II, SPRING HOMEWORK PROBLEMS

Stochastic Modelling in Climate Science

On Mean-Square and Asymptotic Stability for Numerical Approximations of Stochastic Ordinary Differential Equations

Numerical Integration of SDEs: A Short Tutorial

Stochastic optimal control with rough paths

Path Decomposition of Markov Processes. Götz Kersting. University of Frankfurt/Main

Multilevel Monte Carlo for Stochastic McKean-Vlasov Equations

Reflected Brownian Motion

Stochastic Calculus Made Easy

HJB equations. Seminar in Stochastic Modelling in Economics and Finance January 10, 2011

1.1 Definition of BM and its finite-dimensional distributions

Exercise Sheet 8 - Solutions

Moment Properties of Distributions Used in Stochastic Financial Models

An Introduction to Malliavin calculus and its applications

Multi-Factor Lévy Models I: Symmetric alpha-stable (SαS) Lévy Processes

A numerical method for solving uncertain differential equations

Stochastic Calculus February 11, / 33

More Empirical Process Theory

Stochastic Calculus. Kevin Sinclair. August 2, 2016

Weak convergence and large deviation theory

Filtrations, Markov Processes and Martingales. Lectures on Lévy Processes and Stochastic Calculus, Braunschweig, Lecture 3: The Lévy-Itô Decomposition

Exact Simulation of Multivariate Itô Diffusions

for all f satisfying E[ f(x) ] <.

Branching Processes II: Convergence of critical branching to Feller s CSB

Brownian Motion. 1 Definition Brownian Motion Wiener measure... 3

1 Simulating normal (Gaussian) rvs with applications to simulating Brownian motion and geometric Brownian motion in one and two dimensions

Example 4.1 Let X be a random variable and f(t) a given function of time. Then. Y (t) = f(t)x. Y (t) = X sin(ωt + δ)

The Smoluchowski-Kramers Approximation: What model describes a Brownian particle?

FINM 345/Stat 390 Stochastic Calculus, Autumn Lecture 3 (from Singapore) Diffusion Stochastic Calculus

A Class of Fractional Stochastic Differential Equations

Isotonic regression in deterministic and random design settings

PROBABILITY: LIMIT THEOREMS II, SPRING HOMEWORK PROBLEMS

Nonlinear representation, backward SDEs, and application to the Principal-Agent problem

Stochastic Differential Equations in Population Dynamics

Numerical Simulation of Stochastic Differential Equations: Lecture 2, Part 1. Recap: SDE. Euler Maruyama. Lecture 2, Part 1: Euler Maruyama

Nonlinear Techniques for Stochastic Systems of Differential Equations

An adaptive numerical scheme for fractional differential equations with explosions

Let (Ω, F) be a measureable space. A filtration in discrete time is a sequence of. F s F t

Harnack Inequalities and Applications for Stochastic Equations

Transcription:

Lecture 12: Diffusion Processes and Stochastic Differential Equations 1. Diffusion Processes 1.1 Definition of a diffusion process 1.2 Examples 2. Stochastic Differential Equations SDE) 2.1 Stochastic differential equations 2.2 Approximation and simulation for SDE 3. Solving of Stochastic Differential Equations 3.1 Linear SDE 3.2 General linear SDE of diffusion and non-diffusion types 3.3 Transformations of SDE 4. LN Problems 1

1. Diffusion Processes 1.1 Definition of a diffusion process Definition 12.1. A Stochastic real-valued process Xt), t is said to be a diffusion process if it satisfies the following conditions: A: Xt) is a Markov process. B: There exist limits: 1 { µx, t) = lim E Xt + ) Xt)) Xt) = x }, σ 2 1 { x, t) = lim E Xt + ) Xt)) 2 Xt) = x }. C: Xt), t is a continuous process. µx, t) is called a drift coefficient, parameter); σ 2 x, t) is a diffusion coefficient, parameter). 1.2 Examples 1) A Brownian Motion A BM Xt) = X) + µt + σw t), t is a diffusion process with the initial state X) = const, drift µx, t) = µ and diffusion σ 2 x, t) = σ 2. 2

2) A Geometrical Brownian Motion A GBM St) = S)e µt+σw t), t is a diffusion process with the initial state X) = const, µx, t) = xµ + σ2 2 ) and σ 2 x, t) = x 2 σ 2. a) By the definition, µt+s)+σw t+s) St + s) = S)e µs+σw t+s) W t)) = St)e b) Su), u t and W t + s) W t), s are independent. c) Thus St) is a Markov process. Indeed, P{St + s) u St) = v, St k ) = v k, k = 1,..., n} = P{St)e µs+σw t+s) W t)) u St) = v, St k ) = v k, k = 1,..., n} = P{ve µs+σw t+s) W t)) u St) = v, St k ) = v k, k = 1,..., n} = P{ve µs+σw t+s) W t)) u} ) lnu/v) µs = Pt, v, t + s, u) = Φ σ, s where Φu) = 1 2π u e v2 2 dv. d) Also, { 1 } E St + s) St)) St) = x = 1 ) } {x E e µ +σ[w t+ ) W t)] St) 1 = x = x ) e µ + σ2 2 1 = x ) ) 1 + µ + σ2 2 ) + o ) 1 x µ + σ2 2 3 as.

d) Thus, e) Also, µx, t) = xµ + σ2 2 ). { 1 2 } St) E St + ) St)) = x = 1 { ) } 2 St) E x e 2 µ +σw t+ ) W t)) 1 = x { } = x2 E e 2µ +2σZt+ ) Zt)) 2e µ +σw t+ ) W t)) + 1 ) = x2 e 2µ + 4σ2 2 2e µ + σ2 2 + 1 = x2 1 + 2µ + 2σ 2 ) + o ) ) ) 2 1 + µ + σ2 2 ) + o ) + 1 ) = x2 σ 2 + o ) x 2 σ 2 as. f) Thus, σ 2 x, t) = x 2 σ 2. g) St) = S)e µt+σzt) is a continuous process. h) Therefore, the GMB St) ) is a diffusion process with the drift term µx, t) = x µ + σ2 2 and diffusion coefficient σ 2 x, t) = x 2 σ 2. 3). Let W t), t be a standard BM. Consider the process Y t) = W t) 2, t. It is a diffusion process with the initial 4

state Y ) =, the drift coefficient µx, t) 1 and the diffusion coefficient σ 2 x, t) = 4x 2. a) Y t) = W t) 2 is a Markov process. The proof is similar to the proof given in example 2). b) Also, 1 { E W t + ) 2 W t) 2 } W t) 2 = x = 1 {W E t + ) W t) + W t)) 2 W t) 2 } W t) 2 = x = 1 { E W t + ) W t)) 2 + 2W t + ) W t))w t)) } W t) 2 = x = 1 { E W t + ) W t)) 2 + 2W t + ) W t)) x) } W t) 2 = x = 1 E W t + ) W t) ) 2 + 2 xe W t + ) W t) )) = 1 + ) 1. c) Thus, µx, t) 1. 5

d) Also, 1 { W E t + ) 2 W t) 2) 2 } W t) 2 = x W ) 2 ) ) 2 } Zt) = E{ t + ) W t) + 2 W t + ) W t) x 2 = x = 1 W t + ) W t) ) 4 2 E + E4x W t + ) W t) ) 3 + E4x2 W t + ) W t) ) ) 2 = 1 e) Thus, ) 2 3 + 4x + 4x 2 4x 2 as. σ 2 x, t) = 4x 2. f) Y t) is a continuous process. g) Therefore, the process Y t) is a diffusion process with the initial state Y ) =, the drift coefficient µx, t) 1 and the diffusion coefficient σ 2 x, t) = 4x 2. 2. Stochastic Differential Equations SDE) 2.1. Stochastic differential equations 1) A discrete time analogue is a stochastic difference equa- 6

tion: Xn ) Xn 1) ) = µ X n 1) ) ), n 1) + + σ X n 1) ) ) ), n 1) W n ) W n 1) ), n = 1, 2,..., N, where: a) >, N = T ; b) W n ) W n 1) ), n = 1,..., N are independent increments of a standard BM, i.e., independent normal random variables with mean and variance ; c) X) = const R 1. 2) The above stochastic difference equation can be rewritten in the equivalent integral form, Xn ) = X) + n k=1 µ Xk 1) ), k 1) ) + n k=1 X σ k 1) ) ) ), k 1) W k ) W k 1) ), n = 1, 2,..., N, A Continuous time analogues of the above equations are used to define a continuous time stochastic differential equation. 3) Let a real-valued stochastic process Xt), t [, T ] with the initial value X) = const and a standard BM W t), t [, T ] are defined on the same probability space < Ω, F, P > with a filtration F t, t [, T ]. Definition 12.2. We say that the process Xt), t [, T ] 7

satisfies a stochastic differential equation, { dxt) = µ Xt), t ) dt + σ Xt), t ) dw t), t [, T ] 1) iff it satisfies the following stochastic integral equation, { t Xt) = X) + µxs), s)ds + t σxs), s)dw s), t [, T ], 2) which should be understood in the following way: a): Processes Xt) and W t) are adapted to filtration {F t }. b): σ-algebra F t and the process W t + s) W t), s [, T t] are independent, for every t [, T ], c): µx, t) and σx, t) are non-random measurable functions such that stochastic process µxt), t) and σxt), t) belong to the class H 2 [, T ] i.e., T µxt), t) dt and T σ2 Xt), t)dt < 1 with probability 1. d): The stochastic process Xt) has a stochastic differential dxt) = at)dt+bt)dw t) with coefficients at) = µxt), t) and bt) = σxt), t). 4) The first question that should be answered is to give reasonable conditions, under which a solution of equations 1) and 2) exists and is unique. The answer to this and related questions give the following important theorem. Theorem 12.1. Let the coefficients of equation 1) satisfy the following conditions: 8

D: µx, t) and σx, t) satisfy the following inequalities; µx, t) µy, t) + σx, t) σy, t) K x y, µx, t) 2 + σx, t) 2 K 1 + x 2). E: µx, t) and σx, t) are continuous functions in x and t. Then: I: There exists the unique solution of the SDE 1) Xt), t [, T ] for any other solution Xt): P{ Xt) = Xt), t T } = 1). II: Xt) is a diffusion process with coefficients µx, t) and σx, t). III: sup t T EXt) 2 <. a) The proof can be found in SK: Chapter 31. It is based on the use of successive approximation method applied to the stochastic integral equation 2): b) Put X ) t) = X), t [, T ] and then, for n = 1, 2,..., X n) t) = X) + + T T µx n 1) s), s)ds σx n 1) s), s)dw s), t [, T ]. 3) c) Then, it is possible to prove that there exists a process Xt) such that T E X n) t) Xt) 2 dt as n. 4) 9

d) The propositions I III hold for the process Xt). 5) In an obvious way the definition of SDE and Theorem 12.1 can be translated to the interval [t, T ], { dxs) = µxs), s)ds + σxs), s)dw s), 5) s [t, T ]. To indicate the initial value Xt) = x = const one can use the notation X x,t s), s [t, T ] for the solution of the equation 5). Theorem 12.2. Let conditions D and E of Theorem 12.1 hold. Then the diffusion process Xt), which is the solution of equation 1), has transition probabilities, P t, x, s, u) = P{Xs) y/xt) = x} = P{X x,t s) y}, t s. 2.2 Approximation and simulation for SDE 1) One can approximate the SDE 1) by the following stochastic difference equation, X ) n ) X ) n 1) ) = µx ) n 1) ), n 1) ) + σx ) n 1) ), n 1) )W n ) W n 1) )), n = 1, 2,..., N, 6) where: >, N = T. 2) The stochastic difference equation 6) has a recurrence structure that let one solve it by computing sequentially the val- 1

ues X) = X ) ), X ) ),..., X ) N ). 3) There are theorems which let, under conditions analogous to those in Theorem 12.1, interpolate in a reasonable way for example using a linear interpolation) the processes X ) t) between points n, n =, 1,..., N and guarantee appropriate convergence of these interpolated processes to the solution of the SDE 1) as. 4) Note also that the recurrence relation 6) let one effectively simulate the trajectories of the processes X ) t). 3. Solving of Stochastic Differential Equations 3.1 Linear SDE 1) Let us consider a stochastic difference equation, St + ) St) = µst), t) St) + σst), t)w t + ) W t)), 7) where: a) S) = const; b) t =,, 2,..., N. A financial interpretation of this equation is that the rate of return of some pricing process in the interval [t, t + ] is proportional to the length of interval and is given by some normal random variable with mean µst), t) and variance σ 2 St), t) depending on the value of the process at moment t. The assumption is a natural one. 11

2) The continuous tine analogue of equation 7) is a stochastic differential equation dst) St) = µst), t)dt + σst), t)dw t). 8) This classical model was introduced by Samuelson 1965) and Merton 1973). 3) The important is a linear case when takes the form, dst) St) = µt)dt + σt)dw t), 9) where a) S) = const; b) µt) and σt) are non-random continuous functions. 4) Equation 9) can be rewritten as: dst) = St)µt)dt + St)σt)dW t). 1) So, it is a diffusion type SDE with coefficients µx, t) = xµt) and σx, t) = xσt) which are linear functions in x. St) = S)e t µs) 1 2 σ2 s))ds+ t σs)dw s), t [, T ]. 11) Theorem 12.3. The equation 1) has the unique solution which is a diffusion process given explicitly by the following formula, a) Conditions D and E obviously hold and functions µx, t) = xµt) and σx, t) = xσt) are continuous in x and t. So by Theorem 12.1 solution is unique and it is a diffusion process. 12

b) Let us calculate d ln St) using Itô formula. ft, x) = ln x we get, For function f t, f x = 1 x, f xx = 1 x 2 c) Thus, d ln St) = + 1 [ 1 ] St) 2 σt) 2 2 St) 2 + 1 ) St) St)µt) dt + 1 St)σt)dW t), 12) St) or d ln St) = d) This relation is equivalent to ln St) = ln S) + + µt) 1 ) 2 σ2 t) dt + σt)dw t). 13) t e) Relation 14) implies that { t St) = S)e t T. t µs) 1 ) 2 σ2 s) ds σs)dw s), t T. 14) µs) 1 2 σ2 s) ) ds+ t σs)dw s), 15) 5) Note that, by Theorem 12.1, we know that the stochastic process St) given by 13) is: a) diffusion process; b) sup s T ESt) 2 <. 13

6) Equation dst) = St)µdt + St)σdW t) has the unique solution St) = S) exp{µ 1 2 σ2 )t + σw t)} which is a Geometrical Brownian Motion. 3.2 General linear SDE of diffusion and non-diffusion types 1) Method used in the proof of Theorem 12.3 can be applied to linear SDE of non-diffusion type, dst) St) = µt)dt + σt)dw t), t T 16) where S) = const, and we assume that: F: µt) and σt) are continuous stochastic processes such that µt) and σt) belong to H2 [, T ]. G: There is a filtration {F t } such that processes St), µt), σt) and W t) are adapted to the filtration{f t }. H: σ-algebra F t and process W t + s) W t), s T t are independent, for every t [, T ]. Theorem 12.4. Under conditions F H, SDE 16) has the unique solution which is a continuous process given by the following formula, St) = S)e t µs) 1 2 σ2 s))ds+ t σs)dw s), t T. 17) a) Let St) be a process given by 17), i.e, St) = S)e Xt) where Xt) = t µs) 1 2 σs)2 )ds + t σs)dw s). 14

b) The process Xt) has stochastic differential, dxt) = µt) 1 2 σ2 t) ) dt + σt)dw t). c) Let us calculate dfxt)) where fx) = S)e x using Itô formula, f t =, f x = S)e x, f xx = S)e x. d) Thus, dst) = + 1 2 S)eXt) σ 2 t) + S)e Xt) µt) 1 ) 2 σ2 t) dt or +S)e Xt) σt)dw t), dst) = St)µt)dt + St)σt)dW t). e) Therefore, St) = S)e Xt) is a solution of equation 17). f) Let St) be some other continuous solution of equation 17), i.e., d St) = St)µt)dt + St)σt)dW t). g) Then by repeating the calculations given in the proof of Theorem 12.3, we get, then, d ln St) = µt) 1 2 σ2 t) ) dt + σt)dw t), ln St) = ln S) + and finally, t 1 µs) 2 σ2 s) ) t ds + σs)dw s), St) = S)e t µs) 1 2 σ2 s))ds+ t σs)dw s). 15

h) Therefore, St) = St). 3.3 Transformations of SDE 1) Let consider SDE with continuous non-random coefficients µt) and σt), dxt) = Xt)µt)dt + Xt)σt)dW t). Let Y t) = X 2 t) and let us find SDE for Y t). Transformation function is fx) = x 2 that gives us or Thus by Itô formula, f t =, f x = 2x, f xx = 2. dy t) = + 1 2 2 Xt)2 σt) 2 + 2Xt) 2 µt) 2) dt+ +2Xt) Xt)σt)dW t), dy t) =Y t) 2µt) + σ 2 t) ) dt + Y t)2σt)dw t). Both processes Xt) and Y t) are, by Theorem 12.1, diffusion processes that are the unique solutions for the the corresponding SDE. 2) The following theorem presents the transformation method in a general form. Theorem 12.5. Let Xt) is a process, which is the solution of stochastic differential equation, dxt) = µxt), t)dt + σxt), t)dw t), t T. 18) 16

Let also ft, x) is a non-random, monotonic in x for every t, continuous function that has continuous derivatives f t, f x and f x x and an inverse in x for every t function gt, x) ft, gt, x)) = x, gt, ft, x)) = x). Then the process Y t) = ft, Xt)) is a solution of the following SDE, dy t) = [f tt, gt, Y t))) + f xt, gt, Y t)))µt, gt, Y t))) + 1 2 f xxt, gt, Y t)))]dt + f xt, gt, Y t)))σt, gt, Y t)))dw t), t T. 19) a) Using Itô formula we get, dy t) = [f tt, Xt)) + f xt, Xt))µt, Xt)) + 1 2 f xxt, Xt))σt, Xt))]dt + f xt, Xt))σt, Xt))dW t). 2) b) By substituting Xt) = gt, Y t)) in relation 2) we get the SDE 19). 3) If coefficients of SDE 18) and 19) satisfy conditions D and E of Theorem 12.1, then the processes Xt) and Y t) are diffusion processes, which are unique solutions of the SDE, respectively, 18) and 19). In some cases transformation method let one solve SDE explicitly. 4) Let Xt) be a process which has a stochastic differential, dxt) = at)dt + bt)dw t). 21) 17

Relation 21) is actually the simplest SDE which has the solution, Xt) = X) + t as)ds + t 5) Let us now consider SDE of the form, bs)dw s). 22) dxt) = µxt))dt + σxt))dw t), 23) where a) µx) and σx) > are non-random functions that satisfy conditions D and E. Conditions D and E guarantee that there exists a unique solution of SDE 23), which is a diffusion process. 5) Let apply to process Xt) a transformation Y t) = fxt)), where fx) = x dy σy) assuming that integral defining fx) converges for every x. 6) We have, and, by Itô formula, or f t =, f x = 1 σx), f xx = σ x) σx) 2, dy t) = 1 σ Xt)) 2 σxt)) 2 σxt))2 1 + σxt)) µxt))) 1 dt + σxt))dw t) σxt)) dy t) = µxt)) σxt)) 1 ) 2 σ Xt)) dt + dw t). 24) 18

Note that the SDE 23) has the diffusion coefficient σt), while the transformed SDE 24) has the diffusion coefficient 1! 7) Let consider the case, where Then equation 24) takes the form, x µx) σx) 1 2 σ x) = a = const. 25) dy t) = adt + dw t). This equation has explicit solution Y t) = Y ) + at + W t) Let denote f 1 x) the inverse function for function fx) = determined by the relation, dy σy) x = f 1 x) dy σy). Then the solution of the SDE 23) is given by the formula, Example Xt) = f 1 Y t) ) = f 1 y) + at + W t) ). 26) 1) Let σx) = ɛ + x) 2. Then, x dy fx) = ɛ + y) = 1 2 ɛ + y 2) In this case, relation 25) takes the form µx) ɛ + x) 1 2ɛ + x) = a, 2 2 19 x = 1 ɛ 1 ɛ + x. 27)

or µx) = ɛ + x) 3 + aɛ + x) 2. 3) So, initial SDE has the non-linear form, dxt) = ɛ+xt)) 3 +aɛ+xt)) 2 )) ) dt+ɛ+xt)) 2 dw t) 28) 4) In this case, 1 ɛ 1 ɛ + x = y 1 ɛ + x = y + 1 ɛ = 1 ɛy ɛ ɛ + x = ɛ 1 ɛy x = ɛ 1 ɛy ɛ = ɛ2 y 1 ɛy f 1 x) = ɛ2 x 1 ɛx. 5) Thus, the solution of the SDE 28) has the following explicit form, Xt) = ɛ2 X) + at) + W t) ) 1 ɛ X) + at) + W t) ). 29) 4. LN Problems 4.1 Solve SDE dst) St) 4.2 Solve SDE dst) St) = t 2 dt + tdw t). = W t)dt + W t)dw t). 4.3 Let dxt) = e t dt + t 2 dw t). Find SDE for Y t) = Xt) 3. 2

4.4 Let dxt) = µxt))dt + X 3 dw t). Using the transformation Y t) = fxt)), where fx) = x dy y+ɛ), find µx) for 3 which this SDE has an explicit solution and. Find this solution. 4.5 Let dxt) = e t dt + t 3 dzt). Find SDE for Y t) = ln 1 + Xt) ). 4.6 Solve SDE dst) St) = e t dt + W t)dw t). 21