Formalization of a Brownian motion and of stochastic integrals in Lean

9 Local martingales

9.1 Local properties

This section contains material taken mostly from [ and [ .

Definition 9.1

A pre-localizing sequence is a sequence of stopping times \((\tau _n)_{n \in \mathbb {N}}\) such that \(\tau _n \to \infty \) as \(n \to \infty \) (a.s.).

Definition 9.2 Localizing sequence
#

A localizing sequence is a sequence of stopping times \((\tau _n)_{n \in \mathbb {N}}\) such that \(\tau _n\) is non-decreasing and \(\tau _n \to \infty \) as \(n \to \infty \) (a.s.). That is, it is a pre-localizing sequence that is also almost surely non-decreasing.

Lemma 9.3

The constant sequence \(\tau _n = \infty \) is a localizing sequence.

Proof
Lemma 9.4

Let \((\sigma _n), (\tau _n)\) be localizing sequences. Then \((\sigma _n \wedge \tau _n)\) is a localizing sequence.

Proof
Definition 9.5 Local property

Let \(P\) be a class of stochastic processes (or equivalently a predicate on stochastic processes). We say that a stochastic process \(X : T \to \Omega \to E\) is locally in \(P\) (or satisfies \(P\) locally) if there exists a localizing sequence \((\tau _n)_{n \in \mathbb {N}}\) such that for all \(n \in \mathbb {N}\), the process \(X^{\tau _n}\mathbb {I}_{\tau _n {\gt} 0}\) is in \(P\) (in which \(X^{\tau _n}\) denotes the stopped process). We denote the class of processes that are locally in \(P\) by \(P_{\mathrm{loc}}\).

Lemma 9.6

For any class of processes \(P\), we have \(P \subseteq P_{\mathrm{loc}}\).

Proof

Take \(\tau _n = \infty \) for all \(n\).

Lemma 9.7
#

If \(P \subseteq Q\) then \(P_{\mathrm{loc}} \subseteq Q_{\mathrm{loc}}\).

Proof

Let \(X \in P_{\mathrm{loc}}\). Then there exists a localizing sequence \((\tau _n)_{n \in \mathbb {N}}\) such that for all \(n \in \mathbb {N}\), \(X^{\tau _n}\mathbb {I}_{\tau _n {\gt} 0} \in P\). Since \(P \subseteq Q\), for all \(n \in \mathbb {N}\), \(X^{\tau _n}\mathbb {I}_{\tau _n {\gt} 0} \in Q\). Thus \(X \in Q_{\mathrm{loc}}\).

Definition 9.8
#

A class of stochastic processes \(P\) is stable if whenever \(X\) is in \(P\), then for any stopping time \(\tau \), the process \(X^{\tau }\mathbb {I}_{\tau {\gt} 0}\) is also in \(P\).

Lemma 9.9

If \(P\) is a stable class of processes, then \(P_{\mathrm{loc}}\) is also stable.

Proof
Lemma 9.10

If \(P, Q\) are stable classes of processes then \((P\cap Q)_{\mathrm{loc}} = P_{\mathrm{loc}}\cap Q_{\mathrm{loc}}\).

Proof

The forward direction is trivial so we only provide proof for the reverse.

Suppose that \(X \in P_{\mathrm{loc}}\cap Q_{\mathrm{loc}}\). Then, there exists localizing sequences \((\tau _n)_{n \in \mathbb {N}}\) and \((\sigma _n)_{n \in \mathbb {N}}\) such that \(X^{\tau _n} \mathbb {I}_{\tau _n {\gt} 0}\in P\) and \(X^{\sigma _n} \mathbb {I}_{\sigma _n {\gt} 0} \in Q\). Consequently, by the stability of \(P\),

\[ X^{\sigma _n \wedge \tau _n} \mathbb {I}_{\sigma _n \wedge \tau _n {\gt} 0} = (X^{\tau _n} \mathbb {I}_{\tau _n {\gt} 0})^{\sigma _n \wedge \tau _n} \mathbb {I}_{\sigma _n \wedge \tau _n {\gt} 0} \in P. \]

Similarly, by the stability of \(Q\), \(X^{\sigma _n \wedge \tau _n} \mathbb {I}_{\sigma _n \wedge \tau _n {\gt} 0} \in Q\). Thus, as \(\sigma _n \wedge \tau _n\) is a localizing sequence by Lemma 9.4 and \(X^{\sigma _n \wedge \tau _n} \mathbb {I}_{\sigma _n \wedge \tau _n {\gt} 0} \in P \cap Q\) for all \(n\), it follows that \(X \in (P \cap Q)_{\mathrm{loc}}\)

If \((\tau _n)_{n \in \mathbb {N}}\) is a pre-localizing sequence, then the sequence defined by \(\tau '_n = \inf _{m \ge n} \tau _m\) is a localizing sequence.

Proof

Let \(P\) be a stable class of processes and let \((\tau _n)_{n \in \mathbb {N}}\) be a pre-localizing sequence such that for all \(n \in \mathbb {N}\), \(X^{\tau _n}\mathbb {I}_{\tau _n {\gt} 0}\) is in \(P\). If the filtration is right-continuous, then \(X\) is locally in \(P\).

Proof

Using the localizing sequence defined by Lemma 9.11 suffices.

Let \((\tau _n)_{n \in \mathbb {N}}\) be a localizing sequence and let \((\sigma _{n,k})_{k \in \mathbb {N}}\) be a localizing sequence for each \(n\). Then, there exists a strictly increasing sequence \((k_n)_{n \in \mathbb {N}}\) such that the sequence defined by \(\tau '_n = \tau _n \wedge \sigma _{n,k_n}\) is a pre-localizing sequence.

Proof

For each \(n\), since \(\sigma _{n,k} \to \infty \) a.s. as \(k \to \infty \), we may choose \(k_n \in \mathbb {N}\) such that \(P(\sigma _{n,k_n} {\lt} \tau _n \wedge n) \le 2^{-n}\). Then, defining \(\tau '_n = \tau _n \wedge \sigma _{n,k_n}\), we have \(\tau _n' \to \infty \) by the Borel-Cantelli lemma.

Suppose that the filtration is right-continuous. For any stable class of processes \(P\), we have \((P_{\mathrm{loc}})_{\mathrm{loc}} = P_{\mathrm{loc}}\).

Proof

\((P_{\mathrm{loc}})_{\mathrm{loc}} \supseteq P_{\mathrm{loc}}\) by Lemma 9.9 so we only prove the reverse inclusion.

Let \(X\) be a process in \((P_{\mathrm{loc}})_{\mathrm{loc}}\). By definition there exists a localizing sequence \((\tau _n)_{n \in \mathbb {N}}\) such that for all \(n \in \mathbb {N}\), \(X^{\tau _n}\mathbb {I}_{\tau _n {\gt} 0}\) is in \(P_{\mathrm{loc}}\). By definition of \(P_{\mathrm{loc}}\), for each \(n\) there exists a localizing sequence \((\sigma _{n,k})_{k \in \mathbb {N}}\) such that for all \(k \in \mathbb {N}\), \((X^{\tau _n}\mathbb {I}_{\tau _n {\gt} 0})^{\sigma _{n,k}}\mathbb {I}_{\sigma _{n,k} {\gt} 0}\) is in \(P\).

By Lemma 9.12, it suffices to show that there exists a pre-localizing sequence \((\tau '_n)_{n \in \mathbb {N}}\) such that for all \(n \in \mathbb {N}\), \(X^{\tau '_n}\mathbb {I}_{\tau '_n {\gt} 0}\) is in \(P\). Thus, using the localizing sequences \(\tau '_n = \tau _n \wedge \sigma _{n, k_n}\) defined by Lemma 9.13, it remains to argue that by stability of \(P\), \(X^{\tau '_n}\mathbb {I}_{\tau '_n {\gt} 0}\) is in \(P\) for all \(n\). Indeed, this follows as \(X^{\tau '_n}\mathbb {I}_{\tau '_n {\gt} 0} = ((X^{\tau _n}\mathbb {I}_{\tau _n {\gt} 0})^{\sigma _{n,k_n}}\mathbb {I}_{\sigma _{n,k_n} {\gt} 0})^{\tau '_n}\mathbb {I}_{\tau '_n {\gt} 0}\) where \((X^{\tau _n}\mathbb {I}_{\tau _n {\gt} 0})^{\sigma _{n,k_n}}\mathbb {I}_{\sigma _{n,k_n} {\gt} 0}\) is in \(P\) by construction and \(P\) is stable.

Lemma 9.15 Local implication from global implication

Suppose that the filtration is right-continuous. Let \(P, Q\) be two classes of stochastic processes such that \(P \subseteq Q_{\mathrm{loc}}\) and \(Q\) is stable. Let \(X\) be a stochastic process that satisfies \(P\) locally. Then \(X\) satisfies \(Q\) locally. In short, if \(P\) implies \(Q\) locally, then \(P\) locally implies \(Q\) locally.

Proof

Since \(X \in P_{\mathrm{loc}}\), then \(X \in (Q_{\mathrm{loc}})_{\mathrm{loc}}\) by assumption and Lemma 9.7. By Lemma 9.14, \((Q_{\mathrm{loc}})_{\mathrm{loc}} = Q_{\mathrm{loc}}\). Thus \(X \in Q_{\mathrm{loc}}\).

9.2 Locally Cadlag

We in this section assume \(\mathcal{F}\) satisfies the usual conditions (i.e. complete and right-continuous).

Lemma 9.16

Let \(P\) be a predicate on paths and suppose \(X\) is a stochastic process satisfying \(P\) a.s. Then, defining

\[ \tau _n(\omega ) = \begin{cases} \infty & \text{if } X(\omega ) \text{ satisfies } P \\ 0 & \text{otherwise} \end{cases} \]

for all \(n \in \mathbb {N}\), the sequence \((\tau _n)_{n \in \mathbb {N}}\) is a localizing sequence.

Proof
Lemma 9.17

If \(P\) be a predicate on paths such that the constant path \(0\) satisfies \(P\) and \(X\) is a stochastic process satisfying \(P\) a.s. then, \(X\) satisfies \(P\) locally.

Proof

Follows directly by using the localizing sequence defined in Lemma 9.16.

Lemma 9.18

A stochastic process \(X\) is locally right continuous if and only if it is right continuous almost surely.

Proof

If \(X\) is a.s. right continuous, then it is locally right continuous by Lemma 9.17.

On the other hand, assuming \(X\) is locally right continuous, there exists a localizing sequence \((\tau _n)_{n \in \mathbb {N}}\) such that for all \(n \in \mathbb {N}\) and \(\omega \in \Omega \), \((X^{\tau _n}\mathbb {I}_{\tau _n {\gt} 0})(\omega )\) is right continuous. Thus, for almost surely every \(\omega \) and any \(t \in T\) there exists \(N \in \mathbb {N}\) such that \(\tau _N(\omega ) {\gt} t + 1\) (not that the ordering of a.s. and for all is important). Hence, as \(X_s(\omega ) = (X^{\tau _N}\mathbb {I}_{\tau _N {\gt} 0})_s(\omega )\) on a neighborhood of \(t\), we have that \(X(\omega )\) is right continuous at \(t\). Consequently, as \(t\) was arbitrary, \(X\) is a.s. right continuous.

Lemma 9.19

A stochastic process \(X\) has left limits locally if and only if it has left limits almost surely.

Proof

Same proof as in Lemma 9.18.

A stochastic process \(X\) is locally cadlag if and only if it is cadlag almost surely.

Proof

The forward direction follows from Lemmas 9.18 and 9.19 while the reverse direction follows from Lemma 9.17.

Lemma 9.21

The class of right continuous processes is stable.

Proof

Trivial.

Lemma 9.22

The class of processes with left limits is stable.

Proof

Trivial.

Lemma 9.23

The class of cadlag processes is stable.

Proof

Follows from Lemmas 9.21 and 9.22.

9.3 Local martingales

Definition 9.24 Local martingale

We say a stochastic process \((M_t)_{t \in T}\) is a local martingale if it is locally a cadlag martingale in the sense of Definition 9.5. That is, there exists a localizing sequence \((\tau _n)_{n \in \mathbb {N}}\) such that for all \(n \in \mathbb {N}\), the process \(M^{\tau _n}\mathbb {I}_{\tau _n {\gt} 0}\) is a cadlag martingale.

Definition 9.25

A stochastic process is a local submartingale if it is locally a cadlag submartingale in the sense of Definition 9.5. That is, there exists a localizing sequence \((\tau _n)_{n \in \mathbb {N}}\) such that for all \(n \in \mathbb {N}\), the process \(M^{\tau _n}\mathbb {I}_{\tau _n {\gt} 0}\) is a cadlag submartingale.

Lemma 9.26

Every cadlag martingale is a local martingale.

Proof

This follows from Lemma 9.6.

The class of cadlag martingales is stable. That is, if \(M\) is a cadlag martingale and \(\tau \) is a stopping time, then the stopped process cadlag \(M^{\tau }\mathbb {I}_{\tau {\gt} 0}\) is also a martingale.

Proof

Clearly, the stopped process \(M^{\tau }\mathbb {I}_{\tau {\gt} 0}\) is cadlag and it remains to show that it is a martingale.

Fixing \(s \le t \in T\), as \(\{ \tau {\gt} 0\} \in \mathcal{F}_0 \subseteq \mathcal{F}_s\), we have

\[ P[M^{\tau }_t \mathbb {I}_{\tau {\gt} 0} \mid \mathcal{F}_s] = \mathbb {I}_{\tau {\gt} 0}P[M_{\tau \wedge t} \mid \mathcal{F}_{s}]. \]

Thus, as \(\tau \wedge t\) is a bounded stopping time, we have by the optional stopping theorem (Lemma 7.74) that \(P[M_{\tau \wedge t} \mid \mathcal{F}_{s}] = M_{(\tau \wedge t) \wedge s} = M_{\tau \wedge s}\) and so, \(P[M^{\tau }_t \mathbb {I}_{\tau {\gt} 0} \mid \mathcal{F}_s] = M^{\tau }_s \mathbb {I}_{\tau {\gt} 0}\) as required.

The class of cadlag submartingales is stable. That is, if \(M\) is a cadlag submartingale and \(\tau \) is a stopping time, then the stopped process \(M^{\tau }\mathbb {I}_{\tau {\gt} 0}\) is also a cadlag submartingale.

Proof
Theorem 9.29

Let \(M\) be a continuous local martingale with \(M_0 = 0\). If \(M\) is also a finite variation process, then \(M_t = 0\) for all \(t\).

Proof

9.4 Doob-Meyer class

Definition 9.30

We say that a stochastic process is integrable if for all \(t\), \(X_t\) is integrable. A process has locally integrable supremum if \((\sup _{s \le t} \Vert X_s \Vert )_t\) is locally integrable.

Definition 9.31 Doob-Meyer class, class D
#

A stochastic process \((X_t)\) is of class D (or in the Doob-Meyer class) if it is adapted and the set \(\{ X_\tau \mid \tau \text{ is a finite stopping time}\} \) is uniformly integrable.

Definition 9.32 Class DL
#

A stochastic process \((X_t)\) is of class DL if it is adapted and for all \(t \ge 0\), the set \(\{ X_\tau \mid \tau \text{ is a stopping time with } \tau \le t\} \) is uniformly integrable.

A stochastic process of class D is of class DL.

Proof

This follows from the definitions and Lemma 7.55.

Every nonnegative càdlàg submartingale is of class DL.

Proof

Let \(t \in T\) and \(\tau \le t\) be a stopping time. By Lemma 7.75 and nonnegativity we get that \(0 \le X_\tau \le P[X_t \mid X_\tau ]\). Thus, we conclude by applying Lemma 7.51 and Lemma 7.47.

A nonnegative càdlàg submartingale is of class D if and only if it is uniformly integrable.

Proof

Assume that \(X\) is uniformly integrable. We know from Lemma 9.34 that \(X\) is of class DL. Moreover, for any finite stopping time \(\tau \), We have that \(X_\tau = \lim _{n \to +\infty } X_{\tau \land n}\). Thanks to Lemma 7.58, we deduce that \(X\) is of class D.

Conversely, if \(X\) is of class D, then applying Lemma 7.55 using constant stopping times will yield uniform integrability.

Every càdlàg martingale is of class DL.

Proof

Let \(X\) be càdlàg martingale. By Lemma 7.13, \((|X_t|)_{t \in T}\) is a nonnegative càdlàg submartingale, and the result follows from Lemma 9.34 along with Lemma 7.54.

A càdlàg martingale is of class D if and only if it is uniformly integrable.

Proof

Applying Lemma 7.54, this follows from Lemma 9.34 along with Lemma 9.35.

The class of process with locally integrable supremum is stable.

Proof

Let \(X\) be a process with locally integrable supremum and \(\tau \) be a stopping time. Let \(t \in T\). Then \((X^\tau )^*_t = \sup _{s \le t} \| X_{\tau \land s}\| \le \sup _{s \le t} \| X_s\| = X^*_t\), and as \(X^*_t\) is integrable, so is \((X^\tau )^*_t\). Thus \((X^\tau \mathbb {I}_{\tau {\gt} 0})^*_t\) is integrable, concluding the proof.

The class D is stable.

Proof

Let \(X\) be a process of class D and \(\tau \) be a stopping time. For any finite stopping time \(\sigma \), we have that \(X_\sigma ^\tau = X_{\sigma \land \tau }\). Because \(\sigma \land \tau \) is finite and \(X\) is of class D, we deduce from Lemma 7.55 that \(\{ X_{\sigma \land \tau } \mid \sigma \text{ is a finite stopping time}\} \) is uniformly integrable, and thus that \(\{ X^\tau _\sigma \mid \sigma \text{ is a finite stopping time}\} \) is uniformly integrable. Using Lemma 7.51, we obtain that \(\{ X^\tau _\sigma \mathbb {I}_{\tau {\gt} 0} \mid \sigma \text{ is a finite stopping time}\} \) is uniformly integrable, which concludes the proof.

The class DL is stable.

Proof

Let \(X\) be a process of class DL, \(\tau \) be a stopping time. Let \(t \in T\). For any stopping time \(\sigma \le t\), we have that \(X_\sigma ^\tau = X_{\sigma \land \tau }\). Because \(\sigma \land \tau \) is bounded by \(t\) and \(X\) is of class DL, we deduce from Lemma 7.55 that \(\{ X_{\sigma \land \tau } \mid \sigma \text{ is a stopping time with } \sigma \le t\} \) is uniformly integrable, and thus that \(\{ X^\tau _\sigma \mid \sigma \text{ is a stopping time with } \sigma \le t\} \) is uniformly integrable. Using Lemma 7.51, we obtain that \(\{ X^\tau _\sigma \mathbb {I}_{\tau {\gt} 0} \mid \sigma \text{ is a stopping time with } \sigma \le t\} \) is uniformly integrable, which concludes the proof.

Lemma 9.41

Let \(X\) be a stochastic process such that for all \(t \in T\), \(X^*_t := \sup _{s \le t} \| X_t\| \) is integrable. Then \(X\) is of class DL.

Proof

Let \(t \in T\). For every stopping time \(\tau \) with \(\tau \le t\), we have \(\| X_\tau \| \le X^*_t\). Because by hypothesis \(X^*_t\) is integrable, we deduce from Lemma 7.52 that \(\{ X_\tau \mid \tau \text{ is a stopping time with } \tau \le t\} \) is uniformly integrable. This proves that \(X\) is of class DL.

Assume that the filtration is right-continuous. Let \(X\) be a stochastic process with locally integrable supremum. Then \(X\) is locally of class DL.

Proof

Combine Lemma 9.41 and Lemma 9.7.

If \(X\) is of class DL then it is locally of class D.

Proof

Take \(\tau _n := n\). Then

\begin{align*} \{ X^{\tau _n}_\sigma \mid \sigma \text{ is a finite stopping time}\} & = \{ X_{\sigma \land n} \mid \sigma \text{ is a finite stopping time}\} \\ & \subseteq \{ X_\sigma \mid \sigma \text{ is a stopping time with } \sigma \le n\} . \end{align*}

Because \(X\) is of class DL, that last set is uniformly integrable, thus

\[ \{ X^{\tau _n}_\sigma \mid \sigma \text{ is a finite stopping time}\} \]

is uniformly integrable thanks to Lemma 7.55. Lemma 7.51 allows to conclude that

\[ \{ X^{\tau _n}_\sigma \mathbb {I}_{\tau _n {\gt} 0} \mid \sigma \text{ is a finite stopping time}\} \]

is uniformly integrable, thus \(X^{\tau _n} \mathbb {I}_{\tau _n {\gt} 0}\) is of class D. Obviously \(\tau _n \rightarrow +\infty \) as \(n\) goes to infinity, so \(X\) is locally of class D.

If the filtration is right-continuous and \(X\) is locally of class DL then it is locally of class D.

Proof

Apply Lemma 9.15 using Lemma 9.43 and Lemma 9.39.

Lemma 9.45
#

Assume \(T\) is a linear order endowed with a topology making it first countable and \(E\) is a pseudometric space. If \(X\) is a càdlàg process then it maps compact sets to bounded sets.

Proof

Let \(K \subseteq T\) be a compact set and \(\omega \in \Omega \). Assume that \(X(\omega )(K)\) is not bounded. Then there exists a sequence \((t_n)\) in \(K\) such that for all \(n \in N\), \(d(X_{t_n}(\omega ), x) \ge n\), for some arbitrary \(x \in E\). Because \(K\) is compact, there is a subsequence \((t_{\phi (n)})\) that converges. Then one can extract a subsequence \((t_{\phi (\psi (n))})\) which either converges from below or from above. In both cases the sequence \((X_{t_{\phi (\psi (n))}})\) will converge, contradicting the hypotheses.

TODO: refine the hypotheses with those of Début theorem.

Assume \(T\) has a bottom element \(\bot \) and that its closed intervals are compact. If \(X\) is a real-valued càdlàg and adapted process, and if the filtration is right-continuous, then the sequence \(\tau _n := \inf \{ t | X_t \ge n\} \) is a localizing sequence.

Proof

By Corollary 8.4, each \(\tau _n\) is a stopping time. Moreover, for all \(n \in \mathbb {N}\), \(X_t \ge n+1 \implies X_t \ge n\), thus \(\tau _n \le \tau _{n+1}\). Finally, for every \(\omega \in \Omega \) and \(t_0 \in T\) there exists \(N \in \mathbb {N}\) such that for all \(s \le t_0\), \(X_s \le N\) thanks to Lemma ??. Thus for all \(n \ge N\), \(\tau _n(\omega ) \ge t_0\), proving that \(\tau _n\) tends to infinity as n goes to infinity.

For \(Y\) a stochastic process, let \(Y^*_t = \sup _{s \le t} \Vert Y_s \Vert \). Let \(X\) be a stochastic process and let \(\tau = \inf \{ t \mid \Vert X_t \Vert \ge n\} \) for some \(n \in \mathbb {R}\). Then

\begin{align*} (X^{\tau })^*_t \le n + \mathbb {1}_{\tau \le t} \Vert X_{\tau } \Vert \: . \end{align*}
Proof

If \(\tau {\gt} t\), then for all \(s \le t\), \(\| X_s\| \le n\), and thus \((X^\tau )^*_t = \sup _{s \le t} \| X_{\tau \land s}\| \le n = n + \mathbb {1}_{\tau \le t} \| X_{\tau }\| \). Otherwise \((X^\tau )^*_t = \sup _{s \le \tau } \| X_s\| \). For \(s {\lt} \tau \), \(\| X_s\| \le n\), and \(\| X_\tau \| \le \| X_\tau \| \) so \(\sup _{s \le \tau } \| X_s\| \le n \lor \| X_\tau \| \le n + \| X_\tau \| = n + \mathbb {I}_{\tau \le t} \| X_\tau \| \).

Assume \(T\) has a bottom element \(\bot \) and that its closed intervals are compact. If \(X\) is a càdlàg and adapted process, and if the filtration is right-continuous, then any process of class DL has locally integrable supremum.

Proof

Set \(\tau _n := \inf \{ t | X_t \ge n\} \). This is a localizing sequence by Lemma 9.46. For every \(t \in T\), we have by Lemma 9.47 that \((X^{\tau _n})^*_t \le n + \mathbb {I}_{\tau _n \le t} \| X_{\tau _n}\| = n + \mathbb {I}_{\tau _n \le t} \| X_{\tau _n \land t}\| \). Because X is of class DL, \(X_{\tau _n \land t}\) is integrable, so \((X^{\tau _n})^*_t\) is integrable too, so \((X^{\tau _n} \mathbb {I}_{\tau _n {\gt} 0})^*_t\) is integrable. Thus \(X\) has locally integrable supremum.

Assume \(T\) has a bottom element \(\bot \) and that its closed intervals are compact. If \(X\) is a càdlàg and adapted process, and if the filtration is right-continuous, then any process locally of class DL has locally integrable supremum.

Proof

Apply Lemma 9.15 using Lemma 9.48 and Lemma 9.38.

A cadlag adapted process is locally of class D if and only if it has locally integrable supremum.

Proof

A cadlag adapted process is locally of class D if and only if it is locally of class DL.

Proof

Every cadlag submartingale for a right-continuous filtration has locally integrable supremum.

Proof

Define the stopping times \(\sigma _n = \inf \{ t \mid \Vert X_t \Vert \ge n\} \) and set \(\tau _n = \sigma _n \wedge n\). \(\sigma _n\) is a stopping time by Lemma 8.4 and so \(\tau _n\) is also a stopping time. The times \((\tau _n)_{n \in \mathbb {N}}\) form a localizing sequence. By Lemma 9.47, we have that

\begin{align*} (X^{\tau _n})^*_t & \le n + \mathbb {1}_{\tau _n \le t} \Vert X_{\tau _n} \Vert \\ & \le n + \Vert X_{\tau _n} \Vert \: . \end{align*}

It remains to show that \(X_{\tau _n}\) is integrable for each \(n\). This follows by Lemma 7.8 as \(\tau _n\) is a bounded stopping time.

Every cadlag submartingale is locally of class D.

Proof

By Lemma 9.50, it suffices to show that every cadlag submartingale has locally integrable supremum. This is done in Lemma 9.52.

Every local submartingale is locally of class D.

Proof

By Lemma 9.15, it suffices to show that if \(X\) is a submartingale then it is locally of class D. This is done in Lemma 9.53.

Every local martingale is locally of class D.

Proof

A local martingale is a martingale if and only if it is of class DL.

Proof

A nonnegative local submartingale is a submartingale if and only if it is of class DL.

Proof