**Applied Mathematics**

Vol.07 No.08(2016), Article ID:66634,9 pages

10.4236/am.2016.78070

Equivalence of Uniqueness in Law and Joint Uniqueness in Law for SDEs Driven by Poisson Processes

Huiyan Zhao^{1}, Chunhua Hu^{1}, Siyan Xu^{2}

^{1}School of Applied Mathematics, Beijing Normal University Zhuhai, Zhuhai, China

^{2}School of Science, Ningbo University, Ningbo, China

Copyright © 2016 by authors and Scientific Research Publishing Inc.

This work is licensed under the Creative Commons Attribution International License (CC BY).

Received 23 November 2015; accepted 17 May 2016; published 20 May 2016

ABSTRACT

We give an extension result of Watanabe’s characterization for 2-dimensional Poisson processes. By using this result, the equivalence of uniqueness in law and joint uniqueness in law is proved for one-dimensional stochastic differential equations driven by Poisson processes. After that, we give a simplified Engelbert theorem for the stochastic differential equations of this type.

**Keywords:**

Uniqueness in Law, Joint Uniqueness in Law, Poisson Process, Engelbert Theorem

1. Introduction

There are several types of solutions and uniqueness for stochastic differential equations, such as strong solution, weak solution, pathwise uniqueness, uniqueness in law and joint uniqueness in law, which will be introduced in Section 2. The relationship between them was firstly studied by Yamada and Watanabe [1] . They got

and

which is the famous Yamada-Watanabe theorem. It’s an important method to prove the existence of strong solution for SDEs Nowadays. The study on this topic is still alive today and new papers are published, see [2] - [10] . On the other hand, Jacod [11] and Engelbert [12] extended the Yamada-Watanabe theorem to the stochastic differential equation driven by semi-martingales. Especially, Engelbert got an inverse result, that is

which can be seen as a complement of the Yamada-Watanabe theorem. Recently, Kurtz [5] [7] considered an abstract stochastic equation of the form

where, and are Polish spaces. They obtained an unified result ( [7] Theorem 1.5):

which was called the Yamada-Watanabe-Engelbert thereom. This result can cover most results mentioned above. However, joint uniqueness in law is harder to check than uniqueness in law in view of application. The natural question that arises now is: under what conditions, joint uniqueness can be equivalent to uniqueness in law? Kurtz ( [5] [7] ) gave a positive answer for the stochastic equations of the form

when the constrains are simple (linear) equations. It’s sad that the stochastic differential equations are not of the form above, therefore the equivalence does not follow from this result.

There exist few results for this question. As far as we know, Cherny [14] and Brossard [13] proved the equivalence of uniqueness in law and joint uniqueness in law for Itô equations of the following type

driven by Brownian motion with the coefficients which only need to be measurable. Later, Qiao [15] extended the result of [14] to a type of infinite dimensional stochastic differential equaion. For stochastic differential equations with jumps, there is still no such result. So, in this paper, we are concerned with the following one- dimensional stochastic differential equation driven by Poisson process

(1.1)

We will give an extension form of Watanabe’s characterization for 2-dimensional Poisson process, then by applying Cherny’s approach, we prove the equivalence of the uniqueness in law and joint uniqueness in law for Equation (1).

This paper is organized as follows. In Section 2, we prepapre some notations and some definitions. After that, the main results are given and proved in Section 3.

2. Notations and Definitions

Let be the space of all càdlàg functions: and let denote the s-algebra generated by all the maps, , where,. Let.

Definition 2.1. Let be a probability space with a given filtration, and let be a deterministic function of time. A counting process N is a Poisson process with intensity function with respect to the filtration if it satisfies the following conditions.

1) N is adapted to;

2) For all the random variable is independent of;

3) For all, the conditional distribution of the increment is given by

where

Definition 2.2. Let be a probability space with a given filtration, and let, be two deterministic function of time. and are two F-Poisson processes with intensity function and respectively. Process is called a 2-dimensional F-Poisson process with intensity function if and are independent.

We have the following Watanabe characterization for one dimensional Poisson process (see [16] ).

Lemma 2.3. Let be a probability space with a given filtration. Assume that N is a counting process and that is a deterministic function. Assume furthermore that the process M, defined by

is an F-martingale. Then N is a F-Poisson process with intensity function.

In this paper, we consider the following stochastic differential equation driven by the Poisson process

(2.1)

where and are―measurable and for each, and are predictable.

Definition 2.4. A pair, where is a càdlàg -adapted process with paths in and N is a Poisson process with intensity function on a stochastic basis, is called a weak solution of (2.1) if

1) For all,

2) For all,

Definition 2.5. We say that uniqueness in law holds for (2.1) if whenever and are two weak solutions with stochatic bases and such that

then

Definition 2.6. We say that joint uniqueness in law holds for (2.1) if whenever and are two weak solutions with stochatic bases and such that

then

Definition 2.7. We say that pathwise uniqueness holds for (1.1) if whenever and are two weak solutions on the same stochatic bases such that P-a.s., then P-a.s.

3. Main Results

Theorem 3.1. Suppose that the uniqueness in law holds for (2.1). Then, for any solutions and, the law of and the law of are equal on, that is

According to Theorem 1.5 of Kurtz [7] , we have the following simplified Yamada-Watanabe-Engelbert theorem (see aslo [12] Theorem 3, [14] Theorem 3.2) immediately.

Corollary 3.2. The following are equivalent:

1) Equation (2.1) has a strong solution and uniqueness in law holds;

2) Equation (2.1) has a weak solution and pathwise uniqueness holds.

We have the following generalised martingale characterization for 2-dimensional Poisson processes, which may have its own interest.

Lemma 3.3. Let be a probability space with a given filtration. Assume that is a 2-dimensional counting process and that are two deterministic function. Then, N is a 2-dimensional F-Poisson process with intensity function is equivalent to the following two conditions.

1) Processes and defined by

are F-martingales.

2) Process N defined by

is a F―Poisson process with intensity function.

Proof. By Lemma 2.3, we only need to prove that two Poisson processes are independent if and only if their sum is also a Poisson process.

Suppose that and are two independent Poisson processes. For, we have,

By the independence of and, for each, we obtain

We conclude that, which tell us that N is a counting process. Furthermore, we have process defined by

is a martingale. By the Watanabe’s result, we have that N is a Poisson process with intensity function.

On the other hand, suppose that be a Poisson process, we aim to prove that and are independent. In fact, let and, , we have

which completes the proof.

We will recall the concept of conditional distribution from the measure theory. Let be a random element on taking value in a Polish space. Let, then there exists a conditional distribution of with respect to, that is, a family of probability measures on such that

1) For any, the map is -measurable;

2) For any, ,

Remark 3.4. 1) The conditional distribution defined above is unique in the sense: if is another family probability measures with the same properties, then for P-a.e..

2) If is such that, then for P-a.e..

Lemma 3.5. Let be a weak solution of (2.1) on a filtered probability space. Let be a conditional distribution of with respect to (here we consider as a -valued random variable). We denote by Y, M the canonical maps from onto respectively, that is

and

Let

Then, for P-a.e., the pair is a weak solution of (2.1) on.

Proof. Let us check the conditions of Definition 2.4.

1) Firstly, we will check that M is an -Poisson process. For any, , , we have

where is defined as in Definition 2.1. Hence, we have

It follows that

Therefore, for P-a.e.,

We deduce that, for P-a.e, M is an -Poisson process with intensity function.

2) For any,

By Remark 3.4, we have

for P-a.e..

3) We have

Hence,

for P-a.e..

Proof of Theorem 3.1. Let be a weak solution of (2.1) on a filtered space. Let and be two independent -Poisson processes with the same intensity function. Set

Then X, N, and can be defined on in an obvious way. The pair is a solution of (2.1) on and, are independent -Poisson processes. For any, and, is a linear operator from. Let denote the orthogonal projection from; let denote the orthogonal projection from.

For any, set

(3.1)

We claim that and are two independent -Poisson processes with the intensity function. In fact, it’s easy to see that, and are counting processes. Moreover, Let

We have

Note that processes and are predictable precesses and the integrators in the above equations are martingales. We conclude that and are martingale, theorefore is also a martingale. By Lemma 2.3, we get that and are two―Poisson processes with the intensity function and is a―Poisson processes with the intensity function. By Lemma 3.3, we deduce that and are two independent -Poisson processes.

For any, we have

Consequently, is also a solution of (2.1) on.

Let us now consider the filtration

. Note that, for any, the s-fields and are indepen- dent. Thus, is a -Poisson process. So, the pair is also a solution of (2.1) on.

Let be a conditional distribution of with respect to. By Lemma 3.5, is a solution of (2.1) on for -a.e.. As the uniqueness in law holds for (2.1), the distribution (which is the conditional distribution of X with respect to) is the same for -a.e.. This means that the process X is independent of. In particular, X and are independent.

For any and, let be the pseudo inverse of. It is easy to check that is predictable and we have. It follows that

where

By (3.1), we get

The process is a measurable functional of X, while is independent of X. Thus, the last equation shows that the distribution is unique. ,

Remark 3.6. In this paper, the equivalence of the uniqueness in law and joint uniqueness in law holds when diffusion coefficient may be degenerate. We note that, for the general multidimensional stochastic differential equations with jumps, the equivalence does not hold when the diffusion coefficients are allowed to be degenerate. We will consider in the future study.

Acknowledgements

This work was supported in part by the National Natural Science Foundation of China(Grant No.11401029) and Zhejiang Provincial Natural Science Foundation of China (Grant No. LQ13A010020).

Cite this paper

Huiyan Zhao,Chunhua Hu,Siyan Xu, (2016) Equivalence of Uniqueness in Law and Joint Uniqueness in Law for SDEs Driven by Poisson Processes. *Applied Mathematics*,**07**,784-792. doi: 10.4236/am.2016.78070

References

- 1. Yamada, T. and Watanabe, S. (1971) On the Uniqueness of Solutions of Stochastic Differential Equations. Journal of Mathematics-Kyoto University, 11, 155-167.
- 2. Ondrejat, M. (2004) Uniqueness for Stochastic Evolution Equations in Banach Spaces. Dissertationes Mathematicae, 426, 1-63.
- 3. Bahlali, K., Mezerdi, B., N’zi, M. and Ouknine, Y. (2007) Weak Solutions and a Yamada Watanabe Theorem for FBSDEs. Random Operators and Stochastic Equations, 15, 271-286.

http://dx.doi.org/10.1515/rose.2007.016 - 4. Rockner, M.M., Schmuland, B. and Zhang, X. (2008) Yamada-Watanabe Theorem for Stochastic Evolution Equations in Infinite Dimensions. Condensed Matter Physics, 11, 247-259.

http://dx.doi.org/10.5488/CMP.11.2.247 - 5. Kurtz, T.G. (2007) The Yamada-Watanabe-Engelbert Theorem for General Stochastic Equations and Inequalities. Electronic Journal of probability, 12, 951-965.

http://dx.doi.org/10.1214/EJP.v12-431 - 6. Tappe, S. (2013) The Yamada-Watanabe Theorem for Mild Solutions to Stochastic Partial Differential Equations. Electronic Communications in Probability, 18, 1-13.

http://dx.doi.org/10.1214/ECP.v18-2392 - 7. Kurtz, T.G. (2013) Weak and Strong Solutions of General Stochastic Models. arXiv:1305.6747v1
- 8. Graczyk, P. and Malecki, J. (2013) Multidimensional Yamada-Watanabe Theorem and Its Applications to Particle Systems. Journal of Mathematical Physics, 54, 021503.

http://dx.doi.org/10.1063/1.4790507 - 9. Barczy, M., Li, Z. and Pap, G. (2013) Yamada-Watanabe Results for Stochastic Differential Equations with Jumps. arXiv:1312.4485
- 10. Zhao, H. (2014) Yamada-Watanabe Theorem for stochastic Evolution Equation Driven by Poisson Random Measure. ISRN Probability and Statistics, 7 p.
- 11. Jean, J. (1980) Weak and Strong Solutions of Stochastic Differential Equations. Stochastics, 3, 171-191.

http://dx.doi.org/10.1080/17442508008833143 - 12. Engelbert, H. (1991) On the Theorem of T. Yamada and S. Watanabe, Stochastics. An International Journal of Probability and Stochastic Processes, 36, 205-216.

http://dx.doi.org/10.1080/17442509108833718 - 13. Brossard, J. (2003) Deux notions quivalentes dunicit en loi pour les quations diffrentielles stochastiques, Sminaire de Probabilits XXXVII. Springer Berlin Heidelberg, 246-250.
- 14. Cherny, A.S. (2002) On the Uniqueness in Law and the Pathwise Uniqueness for Stochastic Differential Equations. Theory of Probability and Its Applications, 46, 406-419.

http://dx.doi.org/10.1137/S0040585X97979093 - 15. Qiao, H.J. (2010) A Theorem Dual to Yamada-Watanabe Theorem for Stochastic Evolution Equations. Stochastics and Dynamics, 10, 367.
- 16. Watanabe, S. (1964) On Discontinuous Additive Functionals and Lvy Measures of a Markov Process. Japanese Journal of Mathematics, 34, 53-70.