3
$\begingroup$

Can anyone help me with the following SDE?

Solve the following stochastic differential equation: $$dY_t=aY_tdt+(b(t)+cY_t)dB_t$$ with $Y_0=0$.

Hint: Try a solution of the form $Z_tH_t$ where $Z_t = exp(cB_t+(a-\frac{1}{2}c^2t))$ and $dH_t=F(t)dt+G(t)dB_t$ for some adapted process F and G which need to be determined.

Thanks gt6989b for your input!

Please correct me if I'm wrong anywhere.

Since, $$Z_t = exp(cB_t+(a-\frac{1}{2}c^2t))$$ Hence we get: $dZ_t = Z_t(cdB_t-\frac{1}{2}c^2)dt$

and we also have $dH_t=F(t)dt+G(t)dB_t$

Then we apply Ito's Lemma on $Z_tH_t$, which yields \begin{eqnarray*} d(Z_tH_t) &=& Z_tdH_t+H_tdZ_t+dH_tdZ_t \\ &=& Z_t(F(t)dt+G(t)dB_t)+Z_tH_t(cdB_t-\frac{1}{2}c^2dt)+Z_tcG(t)dt \\ &=& Z_t(F(t)-\frac{1}{2}c^2H_t+cG(t))dt+(Z_tG(t)+cZ_tH_t)dB_t \end{eqnarray*}

By letting $Y_t = Z_tH_t$, we compare between the expressions $dY_t$ and $d(Z_tH_t)$ in the $dt$ and $dB_t$ terms respectively. And we get the following:

\begin{eqnarray} Y_t &=& Z_tH_t \\ G(t) &=& \frac{Z_t}{b(t)} \\ F(t) &=& (a+\frac{1}{2}c^2)H_t - c\frac{Z_t}{b(t)} \end{eqnarray}

Note: $F(t) = P$ and $G(t) = Q$ for your P and Q respectively.

But now, I don't really understand how would this result help me in solving the original $dY_t$ SDE.

$\endgroup$
2
  • 1
    $\begingroup$ Well, apply Ito's Lemma to $A_t = Z_t H_t$ that will tell you what $F,G$ need to be to match those of $Y_t$. $\endgroup$
    – gt6989b
    Mar 20, 2013 at 19:50
  • $\begingroup$ what he said ^ follow the hint.... $\endgroup$
    – Lost1
    Mar 20, 2013 at 19:59

2 Answers 2

1
$\begingroup$

Here is an idea to get you started.

Note that $$ dZ_t = cZ_t dB_t + \frac{c^2}{2}Z_t dt = Z_t \left(cdB_t + \frac{c^2}{2}dt\right) $$

and let $dH_t = P dt + Q dB_t$.

Consider $A_t = Z_t H_t$. Then, by Ito's Lemma, and expansion for $dZ_t$,

$$\begin{split} dA_t &= Z_t dH_t + H_t dZ_t + dZ_t dH_t \\ &= Z_t (P dt + Q dB_t) + H_t Z_t \left(cdB_t + \frac{c^2}{2}dt\right) + cQdt \\ &= dt\left[ ??? \right] + dB_t \left[ ??? \right] \end{split} $$

Now match this to $dY_t$, what can you say about $P,Q$?

$\endgroup$
1
  • $\begingroup$ Thx gt6989b, I've updated my question. I hope you can help me in establishing the link between P,Q and the original SDE. Thanks in advance. $\endgroup$
    – Justin K
    Mar 21, 2013 at 0:22
0
$\begingroup$

A linear SDE $$\text{d}Y_{t} = (\alpha(t)+\beta(t)Y_{t})\,\text{d}t+(\gamma(t)+\delta(t)Y_{t})\,\text{d}W_{t}$$ has an explicit (strong) solution which can be found on Wikipedia . Here $\alpha$, $\beta$, $\gamma$ and $\delta$ denote deterministic functions and $W$ a one-dimensional Wiener process. The solution, $Y\ ,$ is expressed in terms of the stochastic exponential, $X\ ,$ which is the strong solution to the linear SDE $$\text{d}X_{t} = \beta(t)Y_{t}\,\text{d}t+\delta(t)Y_{t}\,\text{d}W_{t}\ .$$

$\endgroup$

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.