Brownian Motion and Stochastic Calculus Chapter 1 Section 2

PDF of Solutions

Problem 1.2.2 (not mine)

Let \mathcal C be the collection of subsets A of \Omega such that 1_A(\omega_1) = 1_A(\omega_2).  Note that \mathcal C is a \sigma-algebra (simple to show).  Assume X_s (\omega_1) = X_s (\omega_2) for some s.  Then 1_A (\omega_1)= 1_A(\omega_2) for all A\in \sigma(X_s) so \sigma(X_s) \subset \mathcal C.  Since X_t(\omega_1) = X_t(\omega_2) for all t\in [0, T(\omega_1)] have \sigma(X_{T(\omega_1)}) \subset \mathcal C.  So 1_A(\omega_1) = 1_A(\omega_2) for all A in \sigma(X_{T(\omega_1)}).  Note that B= \{ \omega: T(\omega) = T(\omega_1)\} in \mathcal F_{T(\omega)}^X and 1_B(\omega_1) = 1.  So 1_B (\omega_2) = 1 which implies \omega_2 \in B.  Thus T(\omega_1) = T(\omega_2) as well.

Problem 1.2.6

We wish to show that \{ H_\Gamma < t \} \in \mathcal F_t for all t .  For t=0 , \{H_\Gamma < 0\} = \emptyset so is in \mathcal F_0.  Let t>0.  Consider the set \{ H_\Gamma <t\}.

\{ H_\Gamma < t \} = \{  \omega \in \Omega : H_\Gamma ( \omega) <t \}

=  \{  \omega \in  \Omega : inf ( s \geq 0  : X_s ( \omega) \in \Gamma) < t \}

Let \omega \in \{H_\Gamma < t\}.  By the definition of infimum, there is some s<t such that X_s(\omega) \in \Gamma.  Since \Gamma is open there is some open ball B_\epsilon such that X_s(\omega) \in B_\epsilon \subset \Gamma.   Since X_t(\omega) is right continuous there is some rational number r such that s \leq r < t and X_r(\omega) \in B_\epsilon \subset \Gamma.  

In fact, we can see that \omega \in \{H_\Gamma < t\} if and only if there is some rational number r<t such that X_r(\omega) \in \Gamma.  (We have shown one direction, other direction is trivial).  Thus:

\{H_\Gamma < t\} = \bigcup_{r\in \mathbb Q \cap [0, t)} \{ X_r \in \Gamma\}

If \{ X_r \in \Gamma\} \in \mathcal F_t for each r, we are done (because \mathcal F_t is closed under countable unions).  Note X_t is adapted so \{ X_r \in \Gamma\} \in \mathcal F_r \subset \mathcal F_t.  

Problem 1.2.7

 We wish to show that \{H_\Gamma \leq t\} \in \mathcal F_t for all t.  Begin with t=0.

\{ H_\Gamma \leq 0\} = \{ \omega \in \Omega: X_0(\omega) \in \Gamma\}

= \{X_0 \in \Gamma\}

Since X is adapted, \{X_0 \in \Gamma\} \in \mathcal F_0.   (Does not rely on fact that \Gamma is closed).  

Now assume t>0.  Let us instead show that \{H_\Gamma \leq t\}^C = \{ H_\Gamma > t\} \in \mathcal F_t.   We argue that for closed set \Gamma that  

\{ H_\Gamma > t\} = \bigcap_{s\leq t} \{X_s \in \Gamma^C\}

If \omega \in \{H_\Gamma > t\} then it is certainly true that X_s(\omega) \in \Gamma^C for all s\leq t, so \{ H_\Gamma > t\} \subset \cap\{X_s \in \Gamma^C\}

Now assume \omega \in \cap \{X_s \in \Gamma^C\}.  It is clear that H_\Gamma(\omega) \not < t.  But is it possible that X_t(\omega) \in \Gamma^C and t= H_\Gamma(\omega)?  No.  If t= H_\Gamma(\omega) there are two possibilities.  Either X_t(\omega) \in \Gamma(but have already seen assumed that X_t(\omega) \in \Gamma^C so contradiction).  OR, there is a sequence \{t_n\} such that t\leq t_n, t_n \to t and X_{t_n}(\omega) \in \Gamma for each t_n.  Then by continuity of sample paths X_{t_n}(\omega) \to X_t(\omega).  Thus X_t(\omega) is a limit point of \Gamma.  Since \Gamma is closed, X_t(\omega) \in \Gamma.  Thus, X_t(\omega) \in \Gamma \cap \Gamma^C, again a contradiction.  Hence \cap_{s\leq t} \{X_s \in \Gamma^C\}\subset \{ H_\Gamma > t\} .  

Now we just need to show \cap_{s\leq t} \{X_s \in \Gamma^C\} is in fact a countable union of measurable sets.  This follows directly from continuity of sample paths.  

\bigcap_{s\leq t} \{X_s \in \Gamma^C\}= \bigcap_{r\in \mathbb Q \cap [0,t]} \{X_r \in \Gamma^C\}

And \{X_r \in \Gamma^C\} \in \mathcal F_r \subset \mathcal F_t for each r by the fact that X_t is adapted.

Problem 1.2.10

i). We wish to show that \{ S+T  \leq t \} \in \mathcal F_t .  This is equivalent to showing that \{S + T > t \} \in \mathcal F_t.  Note that :

\{S + T > t \} = \{ S \geq t , T > 0\} \cup \{ S > 0, T \geq t \} \cup \{ S<t, T<t, S+T > t \}

The first two sets on the right hand side are in \mathcal F_t by our preliminary assumptions.  Consider the last set.  

Assume that S,T <t and let \{ q_n \}_{n=1}^\infty be the set of all rational numbers in [0,t).    Note that 

\{ T> q_n \} = \cup_{k=m}^\infty \{T \geq q_n + {1\over k}\}

where m can be chosen to be sufficiently large so that q_n + {1\over m} <t.  Then \{ T> q_n\} \in \mathcal F_t for each q_n and so  \{S\geq t- q_n , T> q_n\}\in \mathcal F_t.

Finally note that 

\{ S,T < t, S+T > t \}  = \cup_{n=1}^\infty  \{ S \geq t- q_n , T > q_n \} \in \mathcal F_t

ii).

\{ S+ T > t\} = \{ T>t\} \cup \{T< t, S\geq t\} \cup \{ S, T <t, S+T > t\}

Recall if T is a stopping time it is also an optional time, so we can use the same work in previous part to show last set on right hand side is in \mathcal F_t .

Problem 1.2.13

Assume A\in \mathcal F_T, i.e A\cap \{T \leq t\} \in \mathcal F_t for all t.  Note 

A^C \cap \{T \leq t\} = \{ T\leq t\} \setminus \{ A\cap \{T \leq t\} \}

The right hand side is in \mathcal F_t be definition of \mathcal F_T.    So A^C \in \mathcal F_T.   Now let A_n \in \mathcal F_T for each n.  Note that 

\bigg{(} \bigcup A_n \bigg{)} \cap \{T \leq t\} = \bigcup (A_n \cap \{ T\leq t\})

Since \mathcal F_t is closed under countable unions, \cup A_n \in \mathcal F_T. So \mathcal F_T is closed under complements and countable unions, so it is a \sigma-algebra.  

Now assume that T(\omega) =t for some fixed t and all \omega\in \Omega. Let A \in \mathcal F_t. Then A\in \mathcal F as well. Note that for \{ T \leq s\} = \emptyset for s< t and \{ T \leq s\} = \Omega for s\geq t. Then, for s <t we have:

A\cap \{T\leq s\} = A\cap \emptyset = \emptyset \in \mathcal F_s

On the other hand, for s\geq t , (noting that \mathcal F_t \subset \mathcal F_s) we have

A \cap \{T\leq s\} = A \cap \Omega = A \in \mathcal F_s

Thus, if A\in \mathcal F_t we see that A\in \mathcal F_T. Trivial to show that \mathcal F_T \subset \mathcal F_t.

Problem 1.2.14

Wish to show that \{ S\leq t\} \in \mathcal F_t for all t.    One equivalent definition of S\in \mathcal F_T is that \{ S\leq s \} \in \mathcal F_T for all s\in \mathbb R.  Note \{ S\leq s \} \in \mathcal F_T by definition means \{S\leq s\} \cap \{T \leq t\} \in \mathcal F_t for all t.  So  \{S\leq s\} \cap \{T \leq t\} \in \mathcal F_t for all s, t.  In particular, let s=t.  Then for all t we have:

\{S \leq t\} \cap \{T \leq t\} \in \mathcal F_t 

But since S\geq T , \{S \leq t\} \cap \{T \leq t\} = \{ S \leq t\}.  Thus, \{S \leq t\} \in \mathcal F_t for all t.  Thus, S is a stopping time.

Problem 1.2.17 (not mine)

i).  Recall definition of restriction to \{ T\leq S\}.  For all A\in \mathcal F , we take A\cap \{T \leq S\}.  Note that since \{T \leq S\} \in \mathcal F_T we have A\cap \{T\leq S\} \in \mathcal F_T for all A\in \mathcal F_T.  Similarly, A\cap \{T\leq S\} \in \mathcal F_S.  Thus A\cap \{T\leq S\} \in \mathcal F_{S\wedge T}.

Want to show that E[ Z| \mathcal F_T] = E[ Z| \mathcal F_{T\wedge S}] so pick a set A in larger \sigma-algebra, \mathcal F_T.  Then for such an A (and on \{T\leq S\}) we have:

E[ 1_A 1_{T\leq S} E[ Z| \mathcal F_T] ] =E[ 1_{A \cap \{T\leq S\}} E[Z|\mathcal F_T] ]

=E[ 1_{A \cap \{T\leq S\}} Z ]

= E[ 1_{A \cap \{T \leq S \} } E[Z| \mathcal F_{T \wedge S } ]  ]

E[ 1_A 1_{T\leq S} E[ Z| \mathcal F_{T\wedge S}] ]

ii).  Similar to previous, we can show that E[Z| \mathcal F_S ] = E[ Z| \mathcal F_{T\wedge S}] on \{S<T\}.  Why ? In previous, didn’t use specific form of inequality (T\leq S versus T<S), just used fact that these sets are in \mathcal F_{T\wedge S}.  

Now to show (ii), will show it on the two disjoint sets \{ T\leq S\} and $\{ S<T\}$.  By (i) we see that

1_{\{ T\leq S\}}E[ E[ Z | \mathcal F_T ] | \mathcal F_{T\wedge S}] = E[ 1_{\{ T\leq S\}} E[ Z | \mathcal F_T ] | \mathcal F_{T\wedge S}]\  \ since\ \ \{ T\leq S\} \in \mathcal F_{T\wedge S}

= E[1_{\{ T\leq S\}} E[ Z| \mathcal F_{T\wedge S} ] | \mathcal F_{T\wedge S} ] \ \ \ \ \ by\ (i)

=1_{\{ T\leq S\}} E[ Z| \mathcal F_{T\wedge S} ]\ \ \ \ \ \ since\ \ \{ T\leq S\} \in \mathcal F_{T\wedge S}

And on \{ S> T\} have something similar, by justification above.


Follow My Blog

Get new content delivered directly to your inbox.