So I am trying to find a closed-form solution for the following SDE.

$$dX_t=a(e^{-t}-X_t)dt+be^{-t}dB \quad X_0=0\quad (I)$$ where $B$ represents a Brownian motion.

My approach, so far, is to use the so-called variation of constant method, i.e. find a suitable $Z=F(X_t,t)$ to be substituted in the original SDE.

Using such an $F$ and Ito's lemma, I can convert the original SDE into $$dZ=\mu_Z(t,Z)dt+\sigma_z(t,z)dB\quad (II)$$


$$\mu_Z=\frac{\partial F}{\partial t}+a(e^{-t}-X_t) \frac{\partial F}{\partial X}+\frac{1}{2}\Big(be^{-t}\Big)^2\frac{\partial^2 F}{\partial X^2}\quad $$


$$\sigma_Z=be^{-t}\frac{\partial F}{\partial X}.$$

Now this is the problem: a "suitable" $F$ must be such that the second SDE (II) turns into another form whose closed-form solution is already known, for example, a GBM.

I have been thinking of several functional forms for $F$ to no avail. Could anyone provide a hint? (Assuming that I am on the right track at all.) Thank you.


1 Answer 1



Let $Y_t := F(X_t, t) := e^{at}(X_t - \frac{a}{1-a}(e^{-at} - e^{-t}))$, if $a \neq 1$, and $e^{t}X_t - t$, if $a = 1$, where $X_t$ is a solution to the above SDE.

  1. Find SDE of $Y_t$ (should be quite nice).
  2. Substitute the integral form of the SDE for $Y_t$ into its definition and rearrange for $X_t$.

The choice of $F$ is not arbitrary. One can derive this using ODE and considering $b=0$, i.e. when there is zero volatility. After solving the ODE, you may notice that the chosen $F(X_t, t)$ (note that $X_t$ is not random under zero volatility assumption) will be constant. The details for this method are outlined here.


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.