Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Consider a symmetric random walk $P(X_i=1)=P(X_i=-1)=1/2$, $S_0=0$, $T_a=\min(n:S_n=a)$

We already know that $P(T_a>T_{-b})=1-P(T_{-b}< T_a)=\frac{b}{a+b}$ and $E(\min\{T_a,T_{-b}\})=ab$.

Prove: $$E(T_a \times 1\{T_a< T_{-b}\})=\frac{(a+2b)ab}{3(a+b)}.$$

I can rewrite the expectation as

$$E(\min\{T_a,T_{-b}\})=E(T_a \times 1\{T_a< T_{-b}\})+E(T_{-b} \times 1\{T_{-b}< T_a\})$$

But I don't know what to do next step? Anyone could give some hints? Thanks a lot.

share|cite|improve this question
I have tried to format your question using $LaTeX$ to make it readable but you might want to check this is what you meant. – Henry Jun 9 '12 at 9:54
Thanks a lot. This is the question i want to ask. – Bonnie Jun 9 '12 at 10:23
up vote 1 down vote accepted

For every integer $-b\leqslant x\leqslant a$, let $t_x=\mathrm E_x(T_a:T_a\lt T_{-b})$. Then $t_{-b}=t_a=0$ and the Markov property after one step yields $$ t_x=\mathrm P_x(T_a\lt T_{-b})+\frac12(t_{x-1}+t_{x+1}), $$ for every integer $-b\lt x\lt a$. Write this as $$ (\Delta t)_x=-\mathrm P_x(T_a\lt T_{-b})=-\frac{x+b}{a+b}, $$ where $\Delta$ is the discrete Laplacian operator, defined by $$ (\Delta u)_x=\frac12(u_{x-1}+u_{x+1})-u_x. $$ Let us check the effect of $\Delta$ on some simple sequences:

  • If $u_x=1$ for every $x$, then $(\Delta u)_x=0$.
  • If $u_x=x$ for every $x$, then $(\Delta u)_x=0$.
  • If $u_x=x^2$ for every $x$, then $(\Delta u)_x=1$.
  • If $u_x=x^3$ for every $x$, then $(\Delta u)_x=3x$.

One sees that $\Delta t=\Delta t^{c,d}$, where, for every $(c,d)$, $t^{c,d}$ is defined by $$ t_x^{c,d}=\frac{c+dx-3bx^2-x^3}{3(a+b)}. $$ If $t_{-b}^{c,d}=t_a^{c,d}=0$, then $t=t^{c,d}$ on $\{-b,a\}$ and $\Delta t=\Delta t^{c,d}$ on $(-b,a)$, thus $\Delta (t-t^{c,d})=0$. The maximum principle shows that $t=t^{c,d}$ everywhere.

In particular, $t_0=\frac{c}{3(a+b)}$ if $(c,d)$ solves the system $t^{c,d}_{-b}=t^{c,d}_a=0$. This yields $c=ab(a+2b)$. QED.

share|cite|improve this answer
Thanks. I'm still a little confused. What's the meaning of $E_x$? – Bonnie Jun 9 '12 at 21:58
$E_x=E(\ \mid S_0=x)$. – Did Jun 10 '12 at 7:39

Your Answer


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.