Stochastic Gronwall inequality explained

Stochastic Gronwall inequality is a generalization of Gronwall's inequality and has been used for proving the well-posedness of path-dependent stochastic differential equations with local monotonicity and coercivity assumption with respect to supremum norm.[1] [2]

Statement

Let

X(t),t\geq0

be a non-negative right-continuous

(l{F}t)t\ge

-adapted process. Assume that

A:[0,infty)\to[0,infty)

is a deterministic non-decreasing càdlàg function with

A(0)=0

and let

H(t),t\geq0

be a non-decreasing and càdlàg adapted process starting from

H(0)\geq0

. Further, let

M(t),t\geq0

be an

(l{F}t)t\ge

- local martingale with

M(0)=0

and càdlàg paths.

Assume that for all

t\geq0

,

X(t)\leq

t
\int
0

X*(u-)dA(u)+M(t)+H(t),

where
*(u):=\sup
X
r\in[0,u]

X(r)

.

and define

c
p=p-p
1-p
. Then the following estimates hold for

p\in(0,1)

and

T>0

:

E(H(T)p)<infty

and

H

is predictable, then

E\left[\left(X*(T)\right)

p\vertl{F}
0\right]\leq
cp
p
p\vertl{F}
E\left[(H(T))
0\right]

\exp\left\lbrace

1/p
c
p

A(T)\right\rbrace

;

E(H(T)p)<infty

and

M

has no negative jumps, then

E\left[\left(X*(T)\right)

p\vertl{F}
0\right]\leq
cp+1
p
p\vertl{F}
E\left[(H(T))
0\right]

\exp\left\lbrace

1/p
(c
p+1)

A(T)\right\rbrace

;

EH(T)<infty,

then

\displaystyle{E\left[\left(X*(T)\right)

p\vertl{F}
0\right]\leq
cp
p

\left(E\left[

p
H(T)\vertl{F}
0\right]\right)

\exp\left\lbrace

1/p
c
p

A(T)\right\rbrace}

;

Proof

It has been proven by Lenglart's inequality.[1]

Notes and References

  1. A stochastic Gronwall lemma and well-posedness of path-dependent SDEs driven by martingale noise . Latin Americal Journal of Probability and Mathematical Statistics. Sima . Mehri. Michael . Scheutzow . 18. 2021. 193–209. 10.30757/ALEA.v18-09. 201660248. free.
  2. Existence and uniqueness of solutions of stochastic functional differential equations. Random Oper. Stoch. Equ.. Max . von Renesse. Michael . Scheutzow . 18. 3. 2010. 267–284. 10.1515/rose.2010.015. 0812.1726 . 18595968.