Scribd
Upload
106 views

Stochastic Calculus and Applications: Example Sheet 4

This document contains 10 problems related to stochastic calculus and applications. Problem 1 involves convergence of random variables under different probability measures. Problem 2 uses Girsanov's theorem to construct a weak solution to an SDE. Problem 3 shows an SDE has strong existence but not pathwise uniqueness.

Uploaded by

Karen
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
106 views

Stochastic Calculus and Applications: Example Sheet 4

This document contains 10 problems related to stochastic calculus and applications. Problem 1 involves convergence of random variables under different probability measures. Problem 2 uses Girsanov's theorem to construct a weak solution to an SDE. Problem 3 shows an SDE has strong existence but not pathwise uniqueness.

Uploaded by

Karen
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

STOCHASTIC CALCULUS AND APPLICATIONS EXAMPLE SHEET 4

Roland Bauerschmidt [email protected], Daniel Heydecker [email protected] Lent 2020


Problems marked with (†) may be handed in for marking (CCA pidgeonhole G/H). Problems marked with (?) are additional questions

Problem 1. Let P and P̃ be probability measures on the same space such that P  P̃.
i. Show that if Zn, Z are random variables such that Zn → Z in P-probability, then Zn → Z in
P̃-probability.
ii. Let X be a continuous semimartingale under both P and P̃. Show that X has the same quadratic
variation process under both measures.

Problem 2. (†) Let b be bounded and measurable. Use Girsanov’s theorem to construct a weak solution
to the SDE
dXt = b(Xt )dt + dBt
over the finite (non-random) time interval [0,T].

Problem 3. (†) Show that the SDE

dXt = 3sign(Xt )|Xt | 1/3 dt + 3|Xt | 2/3 dBt , X0 = 0

has strong existence but not pathwise uniqueness.

Problem 4. Find the unique strong solution to the SDE


1
q
dXt = Xt dt + 1 + Xt2 dBt , X0 = x.
2
(Hint: consider the change of variables Yt = sinh−1 (Xt ).)

Problem 5. (†) Construct a filtered probability space on which a Brownian motion B and an adapted
process X are defined and such that
∫ t
Xs
Xt = ds + Bt , B0 = X0 = 0.
0 s
Is X adapted to the filtration generated by B? Is B a Brownian motion in the filtration generated by X?

Problem 6. Let X be a solution of the SDE

dXt = Xt g(Xt )dBt

where g is bounded and X0 = x > 0 is non-random.


i. By applying Ito’s formula to
 ∫ t
1 t 2
∫ 
Xt exp − g(Xs )dBs + g (Xs )ds
0 2 0

show that P(Xt > 0 for all t ≥ 0) = 1.


ii. Show that E(Xt ) = X0 for all t ≥ 0.
iii. Fix a non-random time horizon T > 0. Show that there exists a measure b P on (Ω, FT ) which is
mutually absolutely continuous with respect to P and a b
P-Brownian motion B
b such that

dYt = Yt g(1/Yt )d B
bt

where Yt = 1/Xt .

1
STOCHASTIC CALCULUS AND APPLICATIONS EXAMPLE SHEET 4

Problem 7. Consider the Cauchy problem for the quasi-linear parabolic equation
∂V 1 1
= ∆V − |∇V | 2 + k on (0, ∞) × Rd ,
∂t 2 2
with V(0, x) = 0 for x ∈ Rd where k : Rd → [0, ∞) is a continuous function. Suppose also that
V : [0, ∞) × Rd → R is continuous on its domain, of class C 1,2 on (0, ∞) × Rd , and satisfies the quadratic
growth condition for every T > 0:
1
−V(t, x) ≤ C + a|x| 2, (t, x) ∈ [0,T] × Rd , a< .
2T
Show that V(t, x) is given by
  ∫ t 
V(t, x) = − log Ex exp − k(Ws )ds
0

for t ≥ 0 and x ∈ Rd .

Problem 8. Let b: Rd → R and σ : Rd → Rd×d be bounded and continuous. For each n, j, set t jn = n2−j
and ψn (t) = t jn if t ∈ [t jn, t j+1
n ). Assume that (X n ) is a tight sequence, and that X n solves
0
∫ t   ∫ t  
Xt = X0 +
n n
b Xψn (u) du +
n
σ Xψnn (u) dBu . (1)
0 0
Show that for each m,T > 0 there exists a constant C > 0 such that
E[|Xtn − Xsn | 2m ] ≤ C(t − s)m for all 0 ≤ s < t ≤ T . (2)
Explain what it means for the sequence (X n ) to be tight in the space C([0,T], Rd ). By looking at the proof
of Kolmogorov’s continuity criterion, explain why (2) implies that (X n ) is tight.

Problem 9. Consider the SDE


dXt = Xt2 dBt . (?)
i. Show that, if (Xt )t ≥0 and (Yt )t ≥0 are two (globally-defined) solutions to (?) with the same starting
point x0 , then they have the same law.
ii. By considering the process X et = 1/|Bt − ξ | where B is a three-dimensional Brownian motion and ξ
3
is a standard Gaussian in R independent of B, show that the SDE has a weak solution.
∫s 2 √
iii. Let Φ(s) = −∞ e−t /2 dt/ 2π be the Gaussian distribution function. Verify that both
 √ 
u1 (t, x) = x 2Φ(1/(x t)) − 1 and u2 (x, t) = x

solve the PDE


∂u x 4 ∂ 2 u
= · , u(0, x) = x on (0, ∞) × (0, ∞).
∂t 2 ∂ x2
iv. Which of these solutions corresponds to u(t, x) = Ex (Xt )?

Problem 10*. Consider the SDE


dXt = −Xt3 dt + dBt ; X0 = x0 . (?)
Recall that there exists a unique maximal solution (Xt )t<ζ to (?).
i. Define T = inf{t ≥ 0 : Xt = 0}. Show that Xt ≤ x0 + Bt for all t ≤ T ∧ ζ and deduce that
Px0 (T0 < ζ) = 1. (3)

ii. Hence show that there exists a sequence of a.s. finite stopping times T0 < S1 < T1 < ... < Sn <
Tn+1 < Sn+1 < .. < ζ such that XTn = 0 and |XSn | = 1 for all n.
iii. Conclude that ζ = ∞ almost surely, so that the solution to (?) is defined for all t ≥ 0.

You might also like