Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

How to calculate the first exit time of the process

$$dx=-M\;dt+dw,$$

where $M$ is a positive constant, and $w$ is a Wiener process? Start from $d>0$, to the boundary $0$.

I solved the differential equation

$$\frac{T''(x)}2-MT'(x)=-1$$

with the condition $T(0)=0$. The solution has two parameters. Using the condition $T(0)=0$, one parameter can be determined. How can we determine the other parameter? Do we need to consider infinity as another boundary, and how to set the boundary condition?

Starting from $d$, this process will almost surely exit via $0$, not $\infty$.

Thanks much for any hint, discussion and help!

share|improve this question
2  
How did you arrive at that differential equation? –  Ben Derrett Nov 23 '11 at 20:24
    
What does it mean to calculate the first exit time? –  Did Nov 23 '11 at 21:47
    
This is a control problem? The original equation is dx=udt+dw. The process starts from 0, and we take u=0 at first. Due to the drift term, suppose the process reaches the boundary d, then we take u=-M to drive the x back to zero. Thus the above differential equation is derived. I want to know the average time to drive x from d to 0. –  Xiangyu Meng Nov 23 '11 at 23:16

1 Answer 1

up vote 5 down vote accepted

Let $\tau=\inf\{t\geqslant0\mid x_t=0\}$. Since $x_t=w_t-Mt+d$, $\tau=\inf\{t\geqslant0\mid w_t=Mt-d\}$. Thus $\tau$ is a stopping time for $(w_t)$. Assume that $\tau$ is integrable. Then the optional stopping theorem yields $\mathrm E(w_\tau)=\mathrm E(w_0)=0$, hence $\mathrm E(x_\tau)=d-M\mathrm E(\tau)$. Since $\tau$ is integrable, $\tau$ is almost surely finite, hence $x_\tau=0$ almost surely, and $\mathrm E(\tau)=d/M$.

It remains to show that $\tau$ is integrable. But $[\tau\geqslant t+(d/M)]\subseteq A_t$ with $A_t=[w_t\geqslant Mt]$. By scaling, $\mathrm P(A_t)=\mathrm P(w_1\geqslant M\sqrt{t})$. When $t\to+\infty$, $\mathrm P(w_1\geqslant M\sqrt{t})=O(\mathrm e^{-t/2})$ hence $\tau$ is integrable (in fact, this proves that $\mathrm e^{c\tau}$ is integrable, for every $c\lt1/2$). The proof is complete.


Another method is to compute $t_h(x)=\mathrm E_x(\tau_{h})$ for every $0\leqslant x\leqslant h$, where $$ \tau_{h}=\inf\{t\geqslant0\mid x_t\in\{0,h\}\}. $$ Then $t_h(0)=t_h(h)=0$ and, as you mentioned, $\frac12t''_h(x)-Mt'_h(x)=-1$ for every $x$ in $(0,h)$. The two boundary conditions yield $$ t_h(x)=\frac{x}M-\frac{h}M\,\frac{\mathrm e^{2Mx}-1}{\mathrm e^{2Mh}-1}. $$ Since $\tau=\sup\limits_h\ \tau_h$, $t_h(x)\to \mathrm E_x(\tau)$ when $h\to+\infty$, and $\mathrm E_x(\tau)=x/M$ for every $x\geqslant0$.

share|improve this answer
    
Thanks very much for your enlightening answers. –  Xiangyu Meng Nov 24 '11 at 17:14

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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