1
$\begingroup$

I would like to obtain an explicit solution to $X$ when it satisfies

$$dX_t = \mu X_t dt + \sigma dW_t, X_S = x$$

Here, $S > 0$, and we want an explicit solution for $X_T$, $T > S$.

I am not sure how to approach the problem. Seems more difficult than regular brownian motion!

$\endgroup$
  • 3
    $\begingroup$ I am voting this question as off-topic for being too basic. Hint 1: Compute the differential of $Y_t = X_t e^{-\mu t}$. Hint 2: This is a special case of an Ornstein-Uhlenbeck process - search a bit and you'll find multiple related questions. $\endgroup$ – LocalVolatility Feb 26 '17 at 14:47
7
$\begingroup$

To solve the SDE you should use the so called variation of constant method. Define a process $Y_t=e^{-\mu t}X_t$, so that using Itô we obtain: $$dY_t=-\mu Y_t dt+ e^{-\mu t}X_t=e^{-\mu t} \sigma dW_t $$ Therefore by integrating we have: $$Y_T=Y_S+\int_S^T e^{-\mu t} \sigma dW_t=e^{-\mu S} X_S +\sigma \int_S^T e^{-\mu t} dW_t$$ $$\Rightarrow \quad X_T=e^{\mu (T-S)} X_S +\sigma \int_S^T e^{\mu (T-t)} dW_t$$ This is a simple Ornstein Uhlenbeck process with mean reversion towards 0 if the coefficient $\mu$ is negative. I think you can easily find it in your Stochastic Calculus reference book.

$\endgroup$
  • $\begingroup$ The integration limits are wrong, no? $\endgroup$ – Jaood Feb 28 '17 at 14:32
  • $\begingroup$ Yes you were right. I fixed it $\endgroup$ – NSZ Mar 1 '17 at 11:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.