getdoce4f9. 168KB Jun 04 2011 12:05:06 AM

Elect. Comm. in Probab. 15 (2010), 213–226

ELECTRONIC
COMMUNICATIONS
in PROBABILITY

STOCHASTIC FLOWS OF DIFFEOMORPHISMS FOR ONE-DIMENSIONAL
SDE WITH DISCONTINUOUS DRIFT
STEFANO ATTANASIO
Scuola Normale Superiore, Piazza dei Cavalieri 7, 56126 Pisa, Italy
email: s.attanasio@sns.it
Submitted July 2,2009, accepted in final form May 24,2010
AMS 2000 Subject classification: 60H10
Keywords: Stochastic flows, Local time
Abstract
The existence of a stochastic flow of class C 1,α , for α < 12 , for a 1-dimensional SDE will be proved
under mild conditions on the regularity of the drift. The diffusion coefficient is assumed constant
for simplicity, while the drift is an autonomous BV function with distributional derivative bounded
from above or from below. To reach this result the continuity of the local time with respect to the
initial datum will also be proved.

1

Introduction

The problem of the existence and smoothness of the stochastic flow under conditions of low regularity of the coefficients has been much studied. Apart from the intrinsic interest of the problem,
there is an interest due to the range of possible applications of these results to PDE theory. For
example, in [3] the uniqueness of the stochastic linear transport equation with Hölder continuous
drift was proved, through new results about stochastic flows of class C 1,α . Because of the greater
regularity of stochastic flows compared to deterministic flows, there are cases in which a PDE admits infinitely many solutions in the deterministic case, but it becomes well posed if it is perturbed
by a stochastic noise. In addition in [2] it is proved that in some cases, through a zero-noise limit,
it is possible to find a criterion to select one particular solution.
We consider an equation of the form


d X tx = b(X tx )d t + dWt
X x (0) = x

(1)

A complete proof of existence of the stochastic flows of class C 1,α is known only when b is Hölder

continuous and bounded, see [3]. In the 1-dimensional case, an important example that deals
with discontinuous b has been studied in [5]. Moreover there are preliminary results in [6]. The
class of bounded variation (BV) fields b emerges from these works as a natural candidate for the
flow property, although only a few partial properties have been proved. Moreover, BV fields are
the most general class considered also in the deterministic literature, see [1]: in any dimension,
213

214

Electronic Communications in Probability

when b ∈ BV and the negative part of the distributional divergence of b is bounded, a generalized
notion of flow exists and is unique.
The aim of this work is to give a complete proof of existence of the stochastic flows of class C 1,α , for
α < 12 , in dimension one, when b ∈ BV and the positive or the negative part of the distributional
derivative of b is bounded. This result, although restricted to the 1-dimensional case, is stronger
than the deterministic one both because we accept a bound on b′ from any side, and because we
construct a flow of class C 1,α , not only a generalized flow.
The partial results of the paper [6] suggest the problem whether b ∈ BV is sufficient. We cannot
reach this result without a one-side control on b′ . The fact that a similar assumption is imposed

in [1] is maybe an indication that it is not possible to avoid it.

2

Flow of homeomorphisms and known results

All results contained in this paper will be proved under the following hypothesis:
1. b ∈ BV (R) and b = b1 − b2 , with b1 and b2 increasing and bounded functions
2. b1 ∈ W 1,∞ or b2 ∈ W 1,∞
We will first suppose b1 ∈ W 1,∞ . Under this hypothesis we will prove that the local time of the
stochastic differential equations (SDE) solutions is Hölder continuous with respect to the initial
data. Thanks to this result we will prove the existence of the stochastic flow of class C 1,α , for
α < 12 . At the end of the paper, using standard facts on the backward equations, we will show that
the results proved hold if we replace the hypothesis b1 ∈ W 1,∞ with the hypothesis b2 ∈ W 1,∞ .
A result about one-dimension stochastic flows, under the hypothesis b ∈ BV, was given in [6].
There it was first proved the non-coalescence property through an elegant proof different from the
one proposed in these notes. Then, the flow continuity was proved, and this property, together
with the continuity of the flow of the backward equation, implies the homeomorphic property of
the flow. However the proof of the continuity of the flow appears to be incomplete. Indeed, in
order to apply Kolmogorov’s lemma, the following inequality is shown:

–
™
E

y

x
sup (X s∧τ
− X s∧τ )2

≤ n(x − y)2

0≤s≤t

where τ is a stopping time depending on x and y. This doesn’t appear sufficient to apply Kolmogorov’s lemma.
Note that, as a standard consequence of the pathwise uniqueness we have that ∀h > 0, a.s.,
X tx+h ≥ X tx . Moreover, assuming b1 ∈ W 1,∞ , we have
Z
(X tx+h

−

X tx ) − (X sx+h

−

X sx )

Z

t

b(X rx+h ) −

=

b(X rx )d r

t

kDb1 k∞ (X rx+h − X rx )d r

≤
s

s

From this inequality, using Gronwall’s lemma we obtain:
(X tx+h − X tx ) ≤ e(t−s)kDb

1

k∞

(X sx+h − X sx )

This fact, together with the proof of the non-coalescence property, contained in [6], is sufficient
to prove the existence of a stochastic flow of homeomorphisms. Therefore the proof of the homeomorphic property contained in [6] can be corrected easily under our hypothesis 1 and 2.

Stochastic flows of diffeomorphisms

215

However we are interested in stronger results about smoothness of the flow. In particular we
are interested in the smoothness of the inverse flow, which is a basic ingredient, for instance, in
the analysis of stochastic transport equations. While the homeomorphic property implies only the
continuity of the inverse flow, we will prove that the inverse flow is of class C 1,α for α < 12 .
Notation
Throughout the paper we will assume given a stochastic basis with a 1-dimensional Brownian
motion (Ω, (F t ) t≥0 , F , P, Wt ). Moreover, for each 0 ≤ s < t we denote by Fs,t the completed σalgebra generated by Wu − Wr for s ≤ r ≤ u ≤ t. We will use the following notation: L = kbk∞ ,
K = kDb1 k∞ , b∗ = b1 + b2 , L ∗ = kb∗ k∞ .
We will denote by X tx the unique solution of the stochastic differential equation (1).

3

Local-time continuity with respect to the initial data

Definition 3.1. Let X tx be the unique solution of equation (1), and let a ∈ R. We will denote by
L ta (X x ) its local time at a, i.e. the continuous and increasing process such that
Z

|X tx

t

sgn(X sx − a)d X sx + L ta (X x )

− a| = |x − a| +
0

Further details about local time can be found in [8].
Remark 1. Recall the following inequality which is used, for example, to prove the continuity with
respect to (a, t) of the local time: Let X t = X 0 + A t + M t be a continuous semimartingale, where
M t is a continuous local martingale, vanishing in 0 and A t is a continuous process with bounded
variation, vanishing in 0. Suppose that sup t≤T |M t | ∨ sup t≤T |A t | ≤ K. Then it holds:
p 
 Z

 t



E 
1{a a and y > x. In the other cases
the following estimates are similar. It holds:
Z
p 

 t∧Tn



x
x
y
y
E sup 
sgn(X s − a)b(X s ) − sgn(X s − b)b(X s )ds 

t≥0  0
p 
Z


 t∧Tn


x
x
x
x
≤ C p E sup 
sgn(X s − a)b(X s ) − sgn(X s − b)b(X s )ds 

t≥0  0


p 
Z
 t∧Tn




+C p E sup 
sgn(X sx − b)b(X sx ) − sgn(X sy − b)b(X sy )ds 


t≥0
0


p 
p 
 Z
 Z
Tn


 Tn





≤ C p E 2L
1a≤X sx 0 such that s ∈ [0, sǫ,x, y (ω)]
ǫ
a
∀a ∈ R, t ∈ [0, T ], and u ∈ [x, y]. This fact will be used in
implies L t+s
(X u ) − L ta (X u ) < 2L ∗ (1+ǫ)
the next theorem, which is crucial to prove the existence of a flow of class C 1,α .
Theorem 4.3. Let x, y, t ∈ R such that x < y, and t ≥ 0. Then, a.s.
Œ
Œ ‚ y
Œ
‚Z
‚Z
X t − X tx
L ta (X u )Db(d a)
≤ sup exp
L ta (X u )Db(d a) ≤
inf exp
u∈[x, y]
y−x
u∈[x, y]
R
R
Proof. Step 1. Fix x < y and t ≥ 0. Fix ǫ > 0, and let sǫ,x, y (ω) be defined as in remark 2.
Moreover define Nǫ,t,x, y (ω) := ⌈ s t (ω) ⌉. Define , ∀ i ∈ N, t i (ω) = (i × sǫ,x, y (ω)) ∧ t. Obviously we
ǫ,x, y

have, a.s., that i ≥ Nǫ,t,x, y (ω) implies t i (ω) = t. Define g : Ω × R+ → R+ as:
g(h) = sup

sup sup |Lsr+a (X z ) − Lsa (X z )|

sup

r∈[0,h] z∈[x, y] s∈[0,t] a∈R

Note that, with the same reasoning used in corollary 3.3 and remark 2, it is possible to show that
g is a.s. continuous, increasing and vanishing in 0. Let z, w ∈ [x, y] such that z < w. Then it
holds:
ΠZ t
‚ w
∞ Z t i+1
X
b(X sw ) − b(X sz )
b(X sw ) − b(X sz )
X t − X tz
ǫ,z,w
ds
=
ds := I0
=
ln
w − Xz
w − Xz
w−z
X
X
s
s
s
s
0
i=0 t
i

Observe that in the last summation a.s. only a finite number of terms are different from 0. Define:
ǫ,z,w
I1

:=

∞
X
i=0

ǫ,z,w
I2

:=

∞
X
i=0

Z

t i+1
ti

1
X twi − X tzi

Z

t i+1

b(X sw ) − b(X sz )
X twi − X tzi

ti

– ‚
b

‚
X sz

+

ds

X twi − X tzi
1+ǫ

ŒŒ

™
−

b(X sz )

ds

220

Electronic Communications in Probability

ǫ,z,w
I3

:=

∞
X

Z

i=0

t i+1
ti

‚

–

1

b

X twi − X tzi

∗

‚
X sz

+

X twi − X tzi

‚

ŒŒ
−b

1−ǫ

∗

‚
X sz

+

X twi − X tzi
1+ǫ

ŒŒ™
ds

Note that the following estimate holds:
ǫ,z,w

I2

ǫ,z,w

− I3

ǫ,z,w

≤ I1

ǫ,z,w

≤ I2

ǫ,z,w

+ I3

(5)

Finally define the random set:



1  w
1  w
Ai,ǫ,z,w := −
X t i − X tzi , −
X t i − X tzi
1−ǫ
1+ǫ
and



A
ρi,ǫ,z,w
(a) := 


1 − ǫ2


 × 1Ai,ǫ,z,w  (a)
2ǫ X twi − X tzi

A
The following properties are immediate: ρi,ǫ,z,w
≥ 0,
1
[− 1−ǫ
(w

R
R

A
ρi,ǫ,z,w
(a)d a = 1, and has support con-

Kt

A
A
tained in
− z)e , 0]. We will use the following notation: ρ̌i,ǫ,z,w
(a) = ρi,ǫ,z,w
(−a).
Similarly we define:

 
1  w
z
X ti − X ti , 0
Bi,ǫ,z,w := −
1+ǫ

and





1+ǫ


B
 × 1Bi,ǫ,z,w  (a)
ρi,ǫ,z,w
(a) :=  
X twi − X tz1
B
A
. In particular its support is contained in
ρi,ǫ,z,w
satisfies properties similar to those of ρi,ǫ,z,w
1
(w − z)e K t , 0].
[− 1+ǫ
Step 2. Thanks to estimates (3) and (4) we have:


∞ Z t i+1
X
b(X sw ) − b(X sz ) b(X sw ) − b(X sz ) 

ǫ,z,w
ǫ,z,w
|I0
− I1 | = 
ds
−
 i=0 t

X sw − X sz
X twi − X tzi
i

≤ǫ

∞
X

Z

i=0

t i+1
ti



 b(X w ) − b(X z ) 

s
s 
 ds

 X tw − X tz

i
i

Using the decomposition b∗ = 2b1 − b, and the relation, which holds for α ≤ β, |b(β) − b(α)| ≤
b∗ (β) − b∗ (α), we obtain:
!
!
Z t i+1
Z t i+1 1 w
∞
∞
X
X
b(X sw ) − b(X sz )
b (X s ) − b1 (X sz )
ǫ,z,w
ǫ,z,w
ds − ǫ
ds
|I0
− I1 | ≤ 2ǫ
X twi − X tzi
X twi − X tzi
t
t
i=0
i=0
i

i

≤

2ǫ
1−ǫ

ǫ,z,w

K t − ǫI1

(6)

Stochastic flows of diffeomorphisms

221

Step 3. From the occupation time formula we have:

‚
‚ w
‚
‚ w
ŒŒ
ŒŒ
∞ Z
X
X t i − X tzi
X t i − X tzi
1

ǫ,z,w
∗
∗
I3
=
−b a+
b a+
 i=0 R (X tw − X tz )
1+ǫ
1−ǫ
i
i


×(L tai+1 (X z ) − L tai (X z ))d a



∞ Z 
X

Š
€
2ǫ


a
z
a
∗
A
z
=
(a)
×
(L
(X
D
b
∗
ρ
)
−
L
(X
))d
a

t i+1
ti
i,ǫ,z,w
 i=0 R 1 − ǫ 2


=

≤


 X
∞ Z 




A
·
z
·
z
∗
ρ̌
∗
(L
(X
)
−
L
(X
))(a)
Db
(d
a)


i,ǫ,z,w
t i+1
ti

1 − ǫ 2  i=0 R
2ǫ

Z h
 X


i

∞
A
·
z
·
z
a
z
a
z

ρ̌
∗
L
(X
)
−
L
(X
)
(a)
−
L
(X
)
−
L
(X
)
i,ǫ,z,w
t
t
t
t

i+1
i
i+1
i
1 − ǫ2
2ǫ

R

i=0

 
Z

2ǫ

×Db (d a) +
L ta (X z )Db∗ (d a)
1 − ǫ2
∗

R


≤



 Z
 


2ǫ
1


z
∗
∗
a
Kt
g
(w − z)e
Nǫ,t,x, y
 L (X )Db (d a) + 4L

1−ǫ
1 − ǫ2  R t
1 − ǫ2
2ǫ

(7)

Step 4. It holds:
ǫ,z,w
I2

=

∞
X
i=0

Z

– ‚

1
R

(X twi − X tzi )
=

∞
X

Z

i=0

=

b

a+

X twi − X tzi
1+ǫ

ŒŒ

™

h
i
− b(a) × L tai+1 (X z ) − L tai (X z ) d a

h
i
Š
€
B
(a) × L tai+1 (X z ) − L tai (X z ) d a
D b ∗ ρi,ǫ,z,w
R

∞ Z h
X
i=0

‚

i

B
L ·t i+1 (X z ) − L ·t i (X z ) ∗ ρ̌i,ǫ,z,w
(a)Db(d a)

R

From this equality it follows:

Z
ǫ,z,w

|I2

L ta (X z )Db(d a)|

−
R



∞ Z nh
X

i
h
io


·
z
·
z
B
a
z
a
z
L t i+1 (X ) − L t i (X ) ∗ ρ̌i,ǫ,z,w (a) − L t i+1 (X ) − L t i (X ) Db(d a)
=
 i=0 R


≤ 4L ∗ g

1
1+ǫ


(w − z)e K t

Nǫ,t,x, y

(8)

222

Electronic Communications in Probability

Step 5. From (5), and from (6), (7) and (8), follows that, assuming ǫ ∈ (0, 12 ), there exists a
constant M dependent only on x, y, t, and on the constants K, L and L ∗ , such that
Z
ǫ,z,w

|I0

L ta (X z )Db(d a)| ≤ M ǫ(1 + sup sup L ta (X u )) + M g(M (w − z))Nǫ,t,x, y

−

(9)

u∈[x, y] a∈R

R

We call ∆ a finite partition of [x, y] if, for some n ∈ N, ∆ = {x = z0 < z1 < · · · < zi < zi+1 < · · · <
zn = y}. Moreover we define |∆| := maxzi ∈∆\ y (zi+1 − zi ). We denote by Λ the set of finite partition
of [x, y]. Obviously it holds:
‚ zi+1
Œ
‚ y
‚ zi+1
z Œ
z Œ
X t − X tx
Xt − Xti
Xt − Xti
≤ inf max
≤
(10)
sup min
∆∈Λ zi ∈∆\ y
zi+1 − zi
y−x
zi+1 − zi
∆∈Λ zi ∈∆\ y
Using (9), we have, ∀ǫ ∈ (0, 21 ) :
‚
inf max

z

z

X t i+1 − X t i
zi+1 − zi

∆∈Λ zi ∈∆\ y

Œ

Œ

‚Z
L ta (X zi )Db(d a)

≤ inf max exp
∆∈Λ zi ∈∆\ y

R

Œ

‚
× exp

M ǫ(1 + sup sup
u∈[x, y] a∈R

Œ

‚Z
L ta (X u )Db(d a)

≤ sup exp
u∈[x, y]

L ta (X u )) +

M g(M |∆|)Nǫ,t,x, y

‚
× exp

Œ
M ǫ(1 + sup sup
u∈[x, y] a∈R

R

L ta (X u ))

With the same reasoning we obtain:
‚ zi+1
Œ
‚Z
z Œ
Xt − Xti
a
z
sup inf
L t (X )Db(d a)
≥ inf exp
z∈[x, y]
zi+1 − zi
∆∈Λ zi ∈∆\ y
R
‚

Œ

× exp −M ǫ(1 + sup sup
u∈[x, y] a∈R

L ta (X u ))

Thanks to the arbitrariness of ǫ, and thanks to relation (10) the proof is complete.
Using the preceding theorem we can now prove the existence of the stochastic flow of class C 1,α .
All results we have proved refers to solutions starting from t = 0. This choice was made to simplify
notations. In the next theorem we will treat the general case. So, we will consider solutions of the
following equation:

s,x
s,x
d X t = b(X t )d t + dWt
(11)
s,x
X (s) = x
0,x

Obviously for X t

theorem 4.3 holds.

Theorem 4.4. Assume conditions 1 and 2 of section 2, and let T > 0. Then there exists a map
(s, t, x, ω) → φs,t (x)(ω) defined for 0 ≤ s ≤ t ≤ T , x ∈ R, ω ∈ Ω with values in R, such that
s,x

s,x

1. given any 0 ≤ s ≤ T , x ∈ R the process X s,x = (X t s ≤ t ≤ T ) defined as X t
continuous Fs,t −measurable solution of equation (11).

= φs,t (x) is a

Stochastic flows of diffeomorphisms

223

−1
−1
2. a.s. φs,t is a diffeomorphism ∀0 ≤ s ≤ t ≤ T and φs,t , φs,t
, Dφs,t , and Dφs,t
are continuous

in (s, t, x), and of class C α in x, for α < 21 .
3. a.s. φs,t (x) = φu,t (φs,u (x)) ∀0 ≤ s ≤ u ≤ t ≤ T and x ∈ R, and φs,s (x) = x.
Moreover an explicit expression for Dφs,t (x) is given by:
‚Z



Dφs,t (x) = exp

−1
L ta (X φ (0,s,x) ) −

−1
Lsa (X φ (0,s,x) )

Œ


Db(d a)

R

Proof. Step 1. We first give the proof under the assumption b1 ∈ W 1,∞ . It will take the first three
steps. Let D be the set of dyadic numbers, and let D T := D ∩ [0, T ]. Define ∀x ∈ D and ∀t ∈ D T ,
0,x
Xet (ω) = X tx (ω). Note the two following facts:
1. Being D a countable set, there exists a negligible set A0 such that, ∀ω 6∈ A0 and ∀x ∈ D X ·x
is a continuous solution of equation (1). In particular since it holds
Z

t

b(X sx )ds + Wt

X tx = x +
0

∀ω 6∈ A0 and x ∈ D, the family of processes {X .x } x∈D is uniformly equicontinuous.
2. Thanks to the countability of the set D × D × D T , there exists a negligible set A1 ⊃ A0 such
that ∀ω 6∈ A1 ∀x, y ∈ D such that x < y, and ∀t ∈ D T it holds
!
Œ
‚Z
‚Z
Œ
0, y
0,x
Xet − Xet
a
u
a
u
L t (X )Db(d a) ≤
inf exp
L t (X )Db(d a)
≤ sup exp
u∈[x, y]
y−x
u∈[x, y]
R
R
These facts imply that, ∀ω 6∈ A1 is well-defined the continuous extension of Xe·0,· (ω) on R × [0, T ].
Denote by φ0,t (x)(ω) this extension. Note that, ∀ω 6∈ A1 , the family {φ0,· (x)(ω)} x∈R is uniformly
equicontinuous; more precisely, ∀ω 6∈ A1 , ∀0 ≤ s ≤ t ≤ T
|φ0,t (x) − φ0,s (x)| ≤ (t − s)L + sup |Wr+(t−s) − Wr |

(12)

r∈[0,T ]

In particular we have that a.s. |φ0,t (x) − x| is bounded in x. This fact, together with continuity
in x, implies that, ∀ω 6∈ A1 , φ0,t (·) is surjective. Moreover it’s immediate to verify that, ∀ω 6∈ A1 ,
and ∀x, y ∈ R, such that x < y, it holds:
Œ ‚
‚Z
Œ
Œ
‚Z
φ0,t ( y) − φ0,t (x)
a
u
a
u
L t (X )Db(d a)
≤ sup exp
L t (X )Db(d a) ≤
inf exp
u∈[x, y]
y−x
u∈[x, y]
R
R
This fact, together with the continuity of L ta (X x ) with respect to (a, t, x) implies that ∀ω 6∈ A1 ,
€R
Š
∀0 ≤ t ≤ T, φ0,t (x)(ω) is differentiable in x, and its derivative is exp R L ta (X x )Db(d a) . So
∀t ∈ [0, T ] a.s. φ0,t is a surjective and differentiable function whose derivative is strictly positive
Š
€R
everywhere, and therefore is a diffeomorphism. Moreover exp R L ta (X x )Db(d a) is of class C α
with respect to x for α < 21 .
Step 2. We now prove that ∀x ∈ R a.s. φ0,t (x) = X tx ∀0 ≤ t ≤ T. Fix x ∈ R. Because of the a.s.
continuity of both φ0,t (x) and X tx with respect to t, and because of the countability of D T , it is

224

Electronic Communications in Probability

sufficient to prove that ∀t ∈ D T , a.s. φ0,t (x) = X tx . So fix t ∈ D T , and let {x n }n∈N be a sequence of
0,x
x
dyadic numbers converging to x. By construction we have that ∀ω 6∈ A1 φ0,t (x n ) = Xet n = X t n and
xn
φ0,t (x) = limn→∞ φ0,t (x n ) = limn→∞ X t . Moreover thanks to theorem 4.3, there is a negligible
set Anx such that ∀ω 6∈ Anx it holds
Œ

‚Z
L ta (X u )Db(d a)

inf exp

u∈[x,x n ]

‚
≤

R

Denote by A x the negligible set

S∞

i
i=0 A x .

x

X t n − X tx

Œ

xn − x

‚Z

Œ
L ta (X u )Db(d a)

≤ sup exp
u∈[x,x n ]

R

Therefore ∀ω 6∈ A x ∪ A1 we have the equality
x

X tx = lim X t n = φ0,t (x)
n→∞

This implies that φ0,t (x) is a continuous F0,t −measurable solution of equation (1).
−1
Step 3. We will denote by φ0,t
(ω) the inverse function of φ0,t (ω). Define ∀0 ≤ s ≤ t ≤ T
−1
φs,t = φ0,t ◦ φ0,s . It’s immediate to verify that property 3 if satisfied. It’s also easy to show that φs,t
−1
−1
is a diffeomorphism, and its inverse function is φs,t
= φ0,s ◦ φ0,t
. Moreover, let {(sn , t n , x n )}n∈N be
a sequence such that 0 ≤ sn ≤ t n ≤ T, x n ∈ R and (sn , t n , x n ) → (s, t, x). Obviously we have
−1
−1
φsn ,t n (x n ) = (φ0,t n ◦ φ0,t
) ◦ φs,t ◦ (φ0,s ◦ φ0,s
(x n ))
n
−1
−1
Thanks to (12), φ0,t n ◦ φ0,t
and φ0,s ◦ φ0,s
converge to the identity; this observation prove
n
−1
that φs,t (x) is jointly continuous in (s, t, x). This implies the continuity of φ0,t
with respect to
−1
(t, x), and thus the continuity of φs,t in (s, t, x). Having proved that the derivative of φ0,t (x) is
€R
Š
exp R L ta (X x )Db(d a) , we have that

‚Z



Dφs,t (x) = exp

−1
L ta (X φ (0,s,x) ) −

−1
Lsa (X φ (0,s,x) )

Œ


Db(d a)

R

−1
Dφs,t
(x)

‚ Z
Œ


a
φ −1 (0,s,x)
a
φ −1 (0,s,x)
= exp −
L t (X
) − Ls (X
) Db(d a)
R

which are jointly continuous in (s, t, x), and of class C α for α < 12 . The property (2) is proved. We
have already proved property (1) if s = 0. To prove property (1) in the general case, fix x ∈ R and
s ≥ 0. Thus applying the Itô formula, we obtain that φs,t (x) is a solution of equation (11). This
fact, and the pathwise uniqueness imply property (1).
Step 4. We give now the proof under the condition b2 ∈ W 1,∞ .
ft := WT −t − WT . Note that W
f is a Brownian motion and the completed σ-algebra generated
Let W
fu − W
fr for s ≤ r ≤ u ≤ t is Fes,t := F T −t,T −s . Note that −b satisfies the conditions used in the
by W
es,t (x), continuous Fes,t −measurable solution of the equation
first three steps. So there exists φ
¨
s,x
s,x
ft
d X t = −b(X t )d t + d W
(13)
s,x
X (s) = x
satisfying property (2) and (3).
e−1
Define, for 0 ≤ s ≤ t ≤ T ψs,t (x)(ω) := φ
T −t,T −s (x)(ω). It’s immediate to verify that ψ satisfies
property (2) and (3). Moreover it’s easy to verify that, for any x ∈ R and 0 ≤ s ≤ T, ψs,· (x)(ω) is
a continuous Fs,t −measurable solution of the equation solution of problem (11).

Stochastic flows of diffeomorphisms

Remark 3. A classic approach to the problem of the existence of a C 1,α stochastic flow is the use
of Itô formula to remove the drift (see [9] and later works on this approach, or a variant in [3]).
Nevertheless, under the assumptions of this paper, there are some difficulties to use this approach.
Indeed, suppose it is possible to prove the existence of a solution of the following parabolic equation
¨
∂ t u(t, x) + 21 ∂ x x u(t, x) + b(x)∂ x u(t, x) = 0
(14)
u(T, x) = x
and that the solution is sufficiently smooth, so that we can apply Itô formula:
1
du(t, X t ) = ∂ t u(t, X t )d t + ∂ x u(t, X t )b(X t )d t + ∂ x u(t, X t )dWt + ∂ x x u(t, X t )d t = ∂ x u(t, X t )dWt
2
Suppose moreover that u(t, ·) is a diffeomorphism. Through the change of variable Yt = u(t, X t )
the original problem has been reduced to the problem of the existence of the flow of the following
stochastic equation
d Yt = σ(t, Yt )dWt
where σ(t, y) = ∂ x u(t, u−1
t ( y)). To solve this problem using well-known theorems, we should prove
σ ∈ C 1,α for some α > 0. But σ ∈ C 1,α implies ∂ x x u(t, ·) ∈ C α . So the discontinuity of the term
b(x)∂ x u(t, x) would be balanced by the term ∂ t u(t, x). However, examples such that ∂ t u(t, x) is
continuous, and b(x) is discontinuous can be shown. Consider, for example, b(x) = sgn(x). It is
t,x
t,x
t,x
possible to prove that u(t, x) = E[X T ] and ∂ t u(t, x) = P(X T < 0) − P(X T > 0). This term, as
α
shown in [4], is continuous in x. Thus we have obtained ∂ x x u(t, ·) 6∈ C . So, even if it is possible
to prove the existence and smoothness of the solution of equation (14), it is not possible to prove
that σ ∈ C 1,α , and it is not possible to prove the existence of the stochastic flow through well-known
theorems.

References
[1] Luigi Ambrosio. Transport equation and Cauchy problem for BV vector fields. Invent. Math.,
158(2):227–260, 2004. MR2096794
[2] Stefano Attanasio and Franco Flandoli. Zero-noise solutions of linear transport equations
without uniqueness: an example. C. R. Math. Acad. Sci. Paris, 347(13-14):753–756, 2009.
MR2543977
[3] F. Flandoli, M. Gubinelli, and E. Priola. Well-posedness of the transport equation by stochastic
perturbation. Invent. Math., 180(1):1–53, 2010. MR2593276
[4] Mihai Gradinaru, Samuel Herrmann, and Bernard Roynette. A singular large deviations phenomenon. Ann. Inst. H. Poincaré Probab. Statist., 37(5):555–580, 2001. MR1851715
[5] Yueyun Hu and Jon Warren. Ray-Knight theorems related to a stochastic flow. Stochastic
Process. Appl., 86(2):287–305, 2000. MR1741809
[6] Youssef Ouknine and Marek Rutkowski. Strong comparison of solutions of one-dimensional
stochastic differential equations. Stochastic Process. Appl., 36(2):217–230, 1990. MR1084976
[7] B. Rajeev. From Tanaka’s formula to Itô’s formula: the fundamental theorem of stochastic
calculus. Proc. Indian Acad. Sci. Math. Sci., 107(3):319–327, 1997. MR1467435

225

226

Electronic Communications in Probability

[8] Daniel Revuz and Marc Yor. Continuous martingales and Brownian motion, volume 293 of
Grundlehren der Mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences]. Springer-Verlag, Berlin, third edition, 1999. MR1725357
[9] A. K. Zvonkin. A transformation of the phase space of a diffusion process that will remove the
drift. Mat. Sb. (N.S.), 93(135):129–149, 152, 1974. MR0336813