Chapter 17

Stochastic integrals: martingale drivers

[8/100%]
Show
LaTeX
Problem 17.1

Show that Proposition 17.2 and Corollary 17.3 remain valid for local martingales with continuous paths (cf. Definition 16.7).

?
Problem 17.2

Extend Theorem 17.1 to local martingales with continuous paths.

?
Problem 17.3

For M,N∈MT, loc cM, N \in \mathcal{M}_{T, \text{ loc }}^{c} we define the covariation by ⟨M,N⟩t:=14(⟨M+N⟩t−⟨M−N⟩t)\langle M, N\rangle_{t}: = \frac{1}{4}\left(\left\langle M+N\right\rangle_{t} - \left\langle M - N\right\rangle_{t}\right). Show that

?
(a)

t↦⟨M,N⟩tt \mapsto \langle M, N\rangle_{t} is well - defined, of bounded variation on compact tt - intervals (BV - process).

(b)

⟨M,N⟩\langle M, N\rangle is the unique BV - process such that MN−⟨M,N⟩∈MT, loc cM N - \langle M, N\rangle \in \mathcal{M}_{T, \text{ loc }}^{c}.

(c)

⟨M,N⟩t=\langle M, N\rangle_{t} = ucp −lim⁡∣Π∣↓0∑Π(Mti∧t−Mti−1∧t)(Nti∧t−Nti−1∧t)- \lim_{\left|\Pi \right| \downarrow 0} \sum_{\Pi }\left(M_{t_{i} \wedge t} - M_{t_{i - 1} \wedge t}\right)\left(N_{t_{i} \wedge t} - N_{t_{i - 1} \wedge t}\right).

(d)

(M,N)↦⟨M,N⟩(M, N) \mapsto \langle M, N\rangle is bilinear.

(e)

∣⟨M,N⟩∣⩽⟨M⟩⟨N⟩\left|\langle M, N\rangle \right| \leqslant \sqrt{\left\langle M\right\rangle } \sqrt{\left\langle N\right\rangle }.

Problem 17.4

Use Lemma 17.14 to show the following assertions:

?
(a)

lim⁡nE[sup⁡t⩽T∣Xtn−Xt∣2]=0⟹Xn→n→∞ ucp X\lim_{n} \mathbb {E}\left[\sup_{t \leqslant T}\left|X_{t}^{n} - X_{t}\right|^{2}\right] = 0 \Longrightarrow X^{n} \xrightarrow [n \rightarrow \infty ]{\text{ ucp }} X.

(b)

Xn→n→∞ ucp X,Yn→n→∞ ucp Y⟹Xn+Yn→n→∞ ucp X+YX^{n} \xrightarrow [n \rightarrow \infty ]{\text{ ucp }} X, \quad Y^{n} \xrightarrow [n \rightarrow \infty ]{\text{ ucp }} Y \Longrightarrow X^{n}+Y^{n} \xrightarrow [n \rightarrow \infty ]{\text{ ucp }} X+Y.

(c)

ET\mathcal{E}_{T} is ucp - dense in LT, loc 2(M)\mathcal{L}_{T, \text{ loc }}^{2}(M) for any M∈MT, loc cM \in \mathcal{M}_{T, \text{ loc }}^{c}.

Problem 17.5

Let M∈MT, loc cM \in \mathcal{M}_{T, \text{ loc }}^{c}. Show that the following maps are continuous, if we use ucpconvergence in both range and domain:

?
(a)

LT, loc 2(M)∋f↦f2⋅⟨M⟩\mathcal{L}_{T, \text{ loc }}^{2}(M) \ni f \mapsto f^{2} \cdot \left\langle M\right\rangle.

(b)

LT, loc 2(M)∋f↦f∙M∈MT, loc c\mathcal{L}_{T, \text{ loc }}^{2}(M) \ni f \mapsto f \bullet M \in \mathcal{M}_{T, \text{ loc }}^{c}.

(c)

MT, loc c∋M↦⟨M⟩\mathcal{M}_{T, \text{ loc }}^{c} \ni M \mapsto \left\langle M\right\rangle.

Problem 17.6

Let M∈MT, loc cM \in \mathcal{M}_{T, \text{ loc }}^{c} and f(t,ω)f(t, \omega ) be an adapted càdlàg (right continuous, finite left limits) process and t∈[0,T]t \in [0, T].

?
(a)

Show that f∈LT0(M)f \in \mathcal{L}_{T}^{0}(M), i.e. ∫0Tf(t,ω)d⟨M⟩t<∞\int_{0}^{T} f(t, \omega ) d\left\langle M\right\rangle_{t} < \infty a.s.

(b)

Show that for any sequence of partitions Π\Pi of [0,T][0, T] with ∣Π∣↓0\left|\Pi \right| \downarrow 0 we have fΠ∙M→f∙Mf^{\Pi } \bullet M \rightarrow f \bullet M in ucp - sense. Here

fΠ(t,ω):=∑Πf(ti−1,ω)  1[ti−1,ti)(t). f^{\Pi }(t, \omega ): = \sum _{\Pi } f\left(t_{i - 1}, \omega \right) \; \mathbb {1}_{\left[t_{i - 1}, t_{i}\right)}(t) .
Problem 17.7

Let M∈MT, loc cM \in \mathcal{M}_{T, \text{ loc }}^{c} and fnf_{n} be a sequence in LT0(M)\mathcal{L}_{T}^{0}(M). If fn→ff_{n} \rightarrow f in ucp - sense, then f∈LT0(M)f \in \mathcal{L}_{T}^{0}(M).

?
Problem 17.8

Show that the process f2∙⟨M⟩t:=∫0t∣f(s)∣2d⟨M⟩sf^{2} \bullet \left\langle M\right\rangle_{t}: = \int_{0}^{t}\left|f(s)\right|^{2} d\left\langle M\right\rangle_{s} appearing in Theorem 17.9.b) is adapted.

?