Loading [MathJax]/jax/output/HTML-CSS/jax.js

Transform a SDE Into a PDE and Vise Versa

Ito's calculus and SDEs
On this page, we learn how to transform a partial differential equation into a stochastic differential equation and vice versa. Dynkin formula and Feynman-Kac formula are defined in this section, but the proofs of these formulae are omitted. Let's denote a random variable Xt starting at x (i.e. X0=x)) by Xxt, and the expectation with respect to Xxt by Ex.

Definition.
Let Xt be a time homogeneous Ito process in Rn such that (Xt+δXt) depends on δ, but not on t. The generator of Xt is then defined by Af(x)=limt0Ex[f(Xt)]f(x)t; xRn. DA denotes the set of functions f:RnR for which the limit exists xRn.

Theorem. Dynkin formula
Let Xt be a time homogeneous Ito process defined as dXt=b(Xt)dt+v(Xt)dBt, and let τ be a stopping time for the process, Ex[τ]<. If fC2(Rn) is a bounded function, and b(Xt) and v(Xt) satisfy bounded conditions , then fDA and Af(x)=ni=1bi(x)fxi+12i,j(vvT)i,j(x)2fxixj, where vT is the transpose of v.

Exercise
Given the following 2dimensional stochastic process dXt=[dX[1]tdX[2]t]=[b1b2]dt+[v1v2v3v4][dB[1]tdB[2]t], where B[1]t and B[2]t are independent Brownian motions.

Find the generator A of the above stochastic process.

Solution.
We haveb=[b1b2]   and   v=[v1v2v3v4] where bi's and vi's are functions of xR. This gives the following partial differential equation Af(x)=b1fx1+b2fx2+12(v21+v22)2fx1x1+(v1v3+v2v4)2fx1x2+12(v23+v24)2fx2x2.

Theorem. Feynman-Kac formula
Let fC2(Rn) be a bounded function and let τ be a stopping time for the process. Assume that qC(Rn) is a lower bounded function and that y(τ,x)=Ex[exp(τ0q(Xz)dz)f(Xτ)].   (1) Then q y+yτ=Ay:τ>0,xRn;   (2)y(0,x)=f(x) :xRn.   (3) If y(τ,x) is bounded on R×Rn and y satisfies the conditions in (2) and (3), then the solution for y is given by the formula in (1).

Let's take an example when the function y(τ,x) is defined over τ[0,T]. We transform the variable τ by t=Tτ and replace x by the security underlying price S. Assume that q is a constant risk-free interest rate with q(St)=r. If a bounded function y(t,St) is defined over [0,T]×[0,), and satisfy r yyt=Ay:t[0,T], St0;y(T,ST)=f(ST) :ST0, where y(T,ST) is the termination value of an option at maturity time T. The solution of y(t,St) is then
y(t,St)=ESt[exp(Tt0rdz)f(STt)]=er(Tt)ESt[f(STt)].


References
  1. Oksendal B., Stochastic Differential Equations: An Introduction With Applications, 5th ed. Springer, 2000.

No comments:

Post a Comment