Research Article | Open Access

Andriy Yurachkivsky, "Convergence of Locally Square Integrable Martingales to a Continuous Local Martingale", *Journal of Probability and Statistics*, vol. 2011, Article ID 580292, 34 pages, 2011. https://doi.org/10.1155/2011/580292

# Convergence of Locally Square Integrable Martingales to a Continuous Local Martingale

**Academic Editor:**Tomasz J. Kozubowski

#### Abstract

Let for each be an -valued locally square integrable martingale w.r.t. a filtration (probability spaces may be different for different ). It is assumed that the discontinuities of are in a sense asymptotically small as and the relation holds for all , row vectors , and bounded uniformly continuous functions . Under these two principal assumptions and a number of technical ones, it is proved that the 's are asymptotically conditionally Gaussian processes with conditionally independent increments. If, moreover, the compound processes converge in distribution to some , then a sequence () converges in distribution to a continuous local martingale with initial value and quadratic characteristic , whose finite-dimensional distributions are explicitly expressed via those of .

#### 1. Introduction

The theory of functional limit theorems for martingales may appear finalized in the monographs [1, 2]. This paper focuses at two points, where the classical results can be refined.

(1) The convergence in distribution to a local martingale with -conditional increments has been studied hitherto in the model, where the -algebra enters the setting along with the prelimit processes. This assumption is worse than restrictiveβit is simply unnatural when one studies the convergence in distribution, not in probability. In the present paper, conditions ensuring asymptotic conditional independence of increments for a sequence of locally square integrable martingales are formulated in terms of quadratic characteristics of the prelimit processes (Theorem 4.5). Our approach to the proving of this property is based on the idea to combine the Stone-Weierstrass theorem (actually its slight modificationβLemma 2.2) with an elementary probabilistic resultβLemma 2.4, which issues in Corollaries 2.7 and 2.8. These corollaries, as well as Lemma 2.4 itself and the cognate Lemma 2.5, will be our tools.

(2) The main object of study in [1, 2] is semimartingale. So, some specific for local martingales facts are passed by. Thus, Theorem VI.6.1 and Corollary VI.6.7 in [2] assert that under appropriate assumptions about semimartingales , the relation where also is a semimartingale, entails the stronger one (below, the notation of convergence in law will be changed). For locally square integrable martingales, one can modify the problem as follows. Let relation (*) be fulfilled. What extra assumptions ensure that is a continuous local martingale and There is neither an answer nor even the question in [1, 2]. A simple set of sufficient conditions is provided by Corollary 5.2 (weaker but not so simple conditions are given by Corollary 5.5). Recalling that the quadratic variation of a continuous local martingale coincides with its quadratic characteristic, we see that the last two relations imply together asymptotic proximity of and . Actually, this conclusion requires even less conditions than in Corollary 5.2. They are listed in Corollary 5.3.

The main results of the paper are, in a sense, converse to Corollaries 5.2 and 5.5. They deal with the problem: what conditions should be adjoined to in order to ensure (**), where is a continuous local martingale with quadratic characteristic ? If the assumptions about the prelimit processes do not guarantee that performing as determines the distribution of , then results of this kind assert existence of convergent subsequences but not convergence of the whole sequence (Theorems 5.1 and 5.4). Combining Theorems 5.4 with 4.5, we obtain Theorem 5.6 asserting that the whole sequence converges to a continuous local martingale whose finite-dimensional distributions are explicitly expressed via those of its initial value and quadratic characteristic. The expression shows that the limiting process has conditionally independent incrementsβbut this conclusion is nothing more than a comment to the theorem.

The proving of the main results needs a lot of preparation. Those technical results which do not deal with the notion of martingale are gathered in Section 2 (excluding Section 2.1), and the more specialized ones are placed in Section 3. The rationale in Sections 3 and 4 would be essentially simpler if we confined ourselves to quasicontinuous processes (for a locally square integrable martingale, this property is tantamount to continuity of its quadratic characteristic). To dispense with this restriction, we use a special technique sketched in Section 2.1.

All vectors are thought of, unless otherwise stated, as columns. The tensor square of will be otherwise written as . We use the Euclidean norm of vectors and the operator norm of matrices. The symbols , and signify: the space of -dimensional row vectors, the class of all symmetric square matrices of a fixed size (in our caseβ) with real entries, and its subclass of nonnegative (in the spectral sense) matrices, respectively.

By , we denote the space of complex-valued bounded continuous functions on a topological space . If and the dimension is determined by the context or does not matter, then we write simply C_{b}.

Our notation of classes of random processes follows [3]. In particular, and signify the class of all martingales with respect to a filtration (= flow of -algebras) and its subclass of uniformly integrable martingales. An -martingale will be called: * square integrable* if for all and* uniformly square integrable* if . The classes of such processes will be denoted and , respectively. The symbol will be suppressed if the filtration either is determined by the context or does not matter. If is a class of -adapted process, then by we denote the respective local class (see [2, Definition I.1.33], where the notation is used). Members of , and are called * local martingales* and * locally* (better * local*) * square integrable martingales*, respectively. All processes, except quadratic variations and quadratic characteristics, are implied -valued, where is chosen arbitrarily and fixed.

The integral will be written shortly (following [1, 2]) as if this integral is pathwise (i.e., is a process of locally bounded variation) or if it is stochastic. We use properties of stochastic integral and other basic facts of stochastic analysis without explanations, relegating the reader to [1β4]. The quadratic variation (see the definition in Section 2.3 [3] or Definition I.4.45 together with Theorem I.4.47 in [2]) of a random process and the quadratic characteristic of will be (and already were) denoted and , respectively. They take values in , which, of course, does not preclude to regard them as -valued random processes.

#### 2. Some Technical Results

The Stone-Weierstrass theorem (see, e.g., [5]) concerns compact spaces only. In the following two, its minor generalizations (for real-valued and complex-valued functions, resp.) both the compactness assumption and the conclusion (that the approximation is uniform on the whole space) are weakened. They are proved likewise their celebrated prototype if one argues for the restrictions of continuous functions to some compact subset fixed beforehand.

Lemma 2.1. *Let be an algebra of real-valued bounded continuous functions on a topological space . Suppose that separates points of and contains the module of each its member and the unity function. Then, for any real-valued bounded continuous function , compact set , and positive number , there exists a function such that and .*

Lemma 2.2. * Let be an algebra of complex-valued bounded continuous functions on a topological space . Suppose that separates points of , and contains the conjugate of each its member and the unity function. Then for any complex-valued bounded continuous function , compact set , and positive number there exists a function such that and .*

We consider henceforth sequences of random processes or random variables given, maybe, on different probability spaces. So, for the th member of a sequence, and should be understood as and . In what follows, βu.i.β means βuniformly integrableβ.

Lemma 2.3. *In order that a sequence of random variables be u.i., it is necessary and sufficient that each its subsequence contain a u.i. subsequence.*

*Proof. *Necessity is obvious; let us prove sufficiency.

Suppose that a sequence is not u.i. Then, there exists such that for all
Consequently, there exists an increasing sequence of natural numbers such that , where . Then, for any infinite set and , we have
which means that the subsequence does not contain u.i. subsequences.

Lemma 2.4. *Let for each be random variables given on a probability space , and a sub--algebra of . Suppose that for each ,
**
and for any the sequence is u.i. Then,
*

*Proof. *Denote . By the second assumption, the sequences ,, are stochastically bounded, which together with the first assumption yields
for any . Hence, writing the identity
and using both assumptions, we get (2.4) for . For arbitrary , this relation is proved by induction whose step coincides, up to notation, with the above argument.

The proof of the next statement is similar.

Lemma 2.5. *Let and be random variables given on a probability space and a sub--algebra of . Suppose that
**
and the sequences , and are u.i. Then,
*

Lemma 2.6. *Let be an -valued random variable given on a probability space , and a sub--algebra of . Suppose that
**
and the relation
**
holds for all from some class of complex-valued bounded continuous functions on which separates points of the latter. Then, it holds for all .*

*Proof. *Let denote the class of all complex-valued bounded continuous functions on satisfying (2.10). Obviously, it is linear. By Lemma 2.4, it contains the product of any two its members. So, is an algebra. By assumption, it separates points of . The other two conditions of Lemma 2.2 are satisfied trivially. Thus, that lemma asserts that for any and , there exists a function such that and . Then,
By the choice of
whence by the dominated convergence theorem
Writing the identity
we get from (2.11)β(2.13)
which together with (2.9) and due to arbitrariness of proves (2.10).

Corollary 2.7. *Let for each be -valued random variables given on a probability space and a sub--algebra of . Suppose that the relations
**
hold for all and from some class of complex-valued bounded continuous functions on which separates points of the latter. Then,
**
for all .*

*Proof. *Denote . Condition (2.17) implies by Lemma 2.4 that relation (2.10) is valid for all of the kind , where . Obviously, such functions separate points of . Furthermore, condition (2.16) where runs over is tantamount to (2.9). It remains to refer to Lemma 2.6.

Corollary 2.8. *Let for each be an -valued random process given on a probability space , a sub--algebra of , and an -measurable -valued random variable. Suppose that the relations
**
and (2.16) hold for , all and any bounded uniformly continuous function on . Then, for any and the relation
**
is valid.*

Recall that for any where is the unit sphere in .

Lemma 2.9. *For any symmetric matrices and ,
*

*Proof. *It suffices to note that the left-hand side of the equality does not exceed .

Let be -valued random processes with trajectories in the Skorokhod space (= cΓ dlΓ g processes on ). We write if the induced by the processes measures on the Borel -algebra in weakly converge to the measure induced by . If herein is continuous, then we write . We say that a sequence is *relatively compact* (r.c.)* in * (*in *) if each its subsequence contains, in turn, a subsequence converging in the respective sense. The weak convergence of finite-dimensional distributions of random processes, in particular the convergence in distribution of random variables, will be denoted . Likewise means equality of distributions.

Denote , Proposition VI.3.26 (items (i), (ii)) [2] together with VI.3.9 [2] asserts that a sequence of cΓ dlΓ g random processes is r.c. in if and only if for all positive and Hence, two consequences are immediate.

Corollary 2.10. *Let and be sequences of -valued and -valued, respectively, cΓ dlΓ g processes such that is r.c. in and for any **
Then, the sequence is also r.c. in C.*

Corollary 2.11. *Let and be r.c. in C sequences of cΓ dlΓ g processes taking values in and , respectively. Suppose also that for each and are given on a common probability space. Then the sequence of -valued processes is also r.c. in C.*

Lemma 2.12. *Let , , and be sequences of cΓ dlΓ g random processes such that for any positive and **
for each **
the sequence is r.c. in . Then, there exists a random process such that .*

*Proof. *Let be a bounded metric in metrizing Skorokhod's -convergence (see, e.g., [2, VI.1.26]). Then, condition (2.26) with arbitrary and implies that
Hence, by the triangle inequality, we have

Let be a uniformly continuous with respect to bounded functional on . Denote , . Then, and for any
which together with (2.29) yields
By condition (2.27),
which jointly with (2.31) proves fundamentality and, therefore, convergence of the sequence . Now, the desired conclusion emerges from relative compactness of in .

Corollary 2.13. *Let the conditions of Lemma 2.12 be fulfilled. Then, , where is the existing by Lemma 2.12 random process such that .*

*Proof. *Repeating the derivation of (2.31) from (2.29), we derive from (2.28) the relation
It remains to write .

Corollary 2.14. *Let , , and be sequences of cΓ dlΓ g random processes such that for any and (2.26) holds; for each relation (2.27) is valid; the sequence is r.c. in C. Then, there exists a random process such that and .*

Below, is the symbol of the locally uniform (i.e., uniform in every interval) convergence.

Lemma 2.15. *Let be cΓ dlΓ g random processes such that . Then, for any -continuous functional on D.*

*Proof. *Lemma VI.1.33 and Corollary VI.1.43 in [2] assert completeness and separability of the metric space , where is the metric used in the proof of Lemma 2.12. Then, it follows from the assumptions of the lemma by Skorokhod's theorem [6] that there exist given on a common probability space cΓ dlΓ g random processes such that (so that is continuous), and a.s. By the choice of , the last relation is tantamount to a.s. Hence, and from continuity of , we get by Proposition VI.1.17 [2] a.s. and, therefore, by the choice of a.s. It remains to note that .

##### 2.1. Forestopping of Random Processes

Let be a filtration on some probability space, an -adapted random process, and a stopping time with respect to . We put and denote ,, . Obviously,
provided exists. In case is -predictable, the operation was called in [7] the * forestopping*. The following three statements were proved in [7].

Lemma 2.16. *Let a random process and a stopping time be -predictable. Then, the process is -predictable.*

Theorem 2.17. *Let be an -martingale and an -predictable stopping time. Then, is a -martingale. If is uniformly integrable, then so is .*

Lemma 2.18. *Let be an -valued right-continuous -predictable random process and a closed set in . Then, the stopping time is -predictable.*

The operation of forestopping was used prior to [7] by Barlow [8] who took the assertion of Theorem 2.17 (which he did not even formulate) for granted.

We will need some subtler properties of this operation.

Lemma 2.19. *Let be a starting from zero locally square integrable martingale with respect to , a positive number, and an -predictable stopping time such that
**
Then, .*

*Proof. *Predictability of implies by Theorem 2.1.13 [3] that there exists a sequence of stopping times such that

By the choice of there exists a sequence of stopping times such that
Then,
Herein, obviously,

From (2.41) we have by Doob's inequality
Noting that: (1) for any , (2) , we may rewrite the last inequality in the form
Writing
we get from (2.37) and (2.38) , which together with (2.45) results in . Then, from (2.42) and (2.43), we have by Fatou's theorem
The assumption yields
Relations (2.38) and (2.39) imply that
which together with (2.47) yields by Fatou's theorem . It remains to note that in view of (2.34).

Lemma 2.20. *Let be a locally square integrable martingale with respect to such that
**
and for any **
Let, further, be a positive number and a predictable time satisfying condition (2.37). Then, .*

*Proof. *In view of (2.50) it suffices to show that . In other words, we may consider that . Then condition (2.51) and the evident inequality
imply by Lemma 2.19 that for any
It remains to prove that for all ,

Taking a sequence with properties (2.40) and (2.41), we write
To deduce (2.54) from this inequality and (2.40), it suffices to note that
so that (2.53) provides uniform integrability of the sequence .

Corollary 2.21. *Under the conditions of Lemma 2.20ββ.*

Theorem 2.22. *Let be a locally square integrable martingale with respect to satisfying conditions (2.50) and (2.51), a positive number, and a predictable time satisfying condition (2.37). Then, .*

*Proof. *Denote . It suffices to show that is a -martingale. To deduce this fact from Theorem 2.17, we note that, firstly, by construction and Lemma 2.20, and, secondly, by construction of both processes and because of (2.35).

#### 3. Martingale Preliminaries

The next statement is obvious.

Lemma 3.1. *Let be a sequence of martingales such that
**
and for any the sequence is uniformly integrable. Then, is a martingale.*

Lemma 3.2. *Let be a sequence of martingales such that (3.1) holds and
**
Then, .*

*Proof. *By condition (3.2) and the definition of quadratic characteristic, there exists a constant such that for all and . Hence, and from (3.1), we have by Fatou's theorem (applicable due to the above-mentioned Skorokhod's principle of common probability space) .

Corollary 3.3. *Let a sequence of square integrable martingales satisfy conditions (3.1) and (3.2). Then, is a uniformly integrable martingale.*

Lemma 3.4. *Let be a local martingale and be an -valued random process. Suppose that they are given on a common probability space and . Then for any a.s.*

*Proof. *By assumption,
for all and . Hence, recalling the definition of quadratic variation, we get .

We shall identify indistinguishable processes, writing simply if . Theorem 2.3.5 [3] asserts that for a continuous local martingale . Hence, and from Lemma 3.4, we have

Corollary 3.5. *Let be a continuous local martingale and an -valued random process. Suppose that they are given on a common probability space and . Then, .*

*Proof. *Lemma 3.4 and formula (3.4) yield . Continuity of both processes enables us to substitute by .

Lemma 3.6. *Let be a locally square integrable martingale. Then, is an increasing process.*

*Proof. *For any , the process is a numeral locally square integrable martingale and, therefore, the process increases. It remains to note that and to recall formula (2.21).

Lemma 3.7. *Let and be locally square integrable martingales with respect to a common filtration. Then,
*

*Proof. *For (then ), this is the Kunita-Watanabe inequality [3, page 118]. In the general case, we take an arbitrary vector and write
hereupon the required conclusion ensues from (2.21) and Lemma 2.9.

Lemma 3.8. *Let and be locally square integrable martingales with respect to some common filtration. Then, for any *

*Proof. *Writing the identities
we deduce from Lemma 3.7 that
It remains to note that the right-hand side increases in by Lemma 3.6.

For a function we denote .

Let us introduce the conditions:(RC) The sequence is r.c. in .(UI1) The sequence is u.i.(UI2) For any the sequence is u.i.(UI3) The sequence is u.i.

Lemma 3.9. *Let be a sequence of local square integrable martingales satisfying the conditions: (RC),
**
and, for each , the condition
**
Then, is r.c. in C.*

*Proof. *It follows from (RC) and (3.10) by Rebolledo's theorem [2, VI.4.13] that is r.c. in . Hereon, the desired conclusion follows from Proposition VI.3.26 (items (i) and (iii)) [2] with account of VI.3.9 [2].

Combining Lemma 3.9 with Corollary 2.11, we get

Corollary 3.10. *Under the assumptions of Lemma 3.9, the sequence of compound processes is r.c. in C.*

Some statements below deal with random processes on , not on . In this case, the time variable is denoted by and C means instead of .

Lemma 3.11. *Let be a r.c. in C and satisfying condition (UI1) sequence of martingales on . Then, for any , the sequence is u.i.*

*Proof. *The obvious equality allows us to consider that . Then, condition (UI1) together with Doob's inequality yields
whence

By assumption, for any infinite set , there exist an infinite subset and a random process such that
Condition (UI1) implies that is a square integrable martingale and for any ,

From (3.14) and (3.13), we have by Corollary VI.6.7 [2]
Hence, and from (3.15), recalling that for any -valued one has [3, Theorem ], we get
Comparing this relation with (3.16), we conclude that the sequence is uniformly integrable. Hence, in view of arbitrariness of , we deduce by Lemma 2.3 uniform integrability of .

Lemma 3.12. *Let be a sequence of martingales on satisfying condition (UI1) and (UI2). Suppose that there exists an -valued random process such that
**
Then, firstly,
**
where is the null matrix, is a continuous martingale, and, secondly,
*

*Proof. *For the same reason as in the proof of Lemma 3.11, we may consider that . Then, as was shown above, condition (UI1) implies (3.13). Combining the latter with
(a part of (3.18)), we get by Corollary VI.6.7 [2] that
From (3.18) and (3.22), we get by Corollary 2.11 that for any infinite set , there exist an infinite subset and an -valued random process such that
(Of course .)

Denote . This is a martingale by Lemma 10.4 in [4]. Relation (3.23) implies that
For any , the sequence is, by Lemma 3.11 and condition (UI2), u.i. So, relation (3.24) implies by Lemma 3.1 that is a martingale. Also, it implies its continuity. Relation (3.23) shows that the processes and increase and start from zero. So, starts from zero and has finite variation in . These four properties of imply together that for all . Thus, any subsequence contains, in turn, a subsequence such that . This proves (3.19).

From (3.22) and (3.19), we have . And this is, in view of (3.4), tantamount to
Comparing this relation with (3.18), we arrive at (3.20).

*Remark 3.13. **The second conclusion of Lemma 3.12 implies by Corollary 3.5 that *.

Corollary 3.14. *Let a sequence of martingales on satisfy conditions (RC), (3.11), (UI1), and (UI2). Then, relation (3.19) holds.*

*Proof. *By Corollary 3.10 for any infinite set , there exist an infinite set and an -valued random process such that relation (3.18) holds as . Then, by Lemma 3.12, so does (as ) (3.19). Due to arbitrariness of this relation holds when ranges over , too.

Corollary 3.15. *Let be a sequence of martingales on satisfying conditions (RC) and for all , conditions (3.11), (UI1), (UI2). Then, relation (3.19) holds.*

Lemma 3.16. *Let a sequence of martingales on satisfy conditions (RC) and, for any , (3.11) and (UI3). Then, relation (3.19) holds.*

*Proof. *Denote . Obviously,
By Corollary 2.21ββ is a square integrable martingale with respect to . By Theorem 2.22,
Condition (RC) implies relative compactness of the sequence . By construction . So, condition (UI3) implies that for any and the sequence is u.i. Thus, Corollary 3.15 asserts that for any
Equalities (3.27) and (2.36) yield the relation
which together with (3.26), (3.28) and (RC) entails (3.19).

Corollary 3.15 and Lemma 3.16 are only the steps towards the final result about asymptotic proximity of quadratic variations and quadratic characteristicsβCorollary 5.3.

Corollary 3.17. *Let a sequence of martingales on satisfy conditions (RC) and for any , (3.11) and (UI3). Suppose also that there exists an -valued random process such that relation (3.18) is valid. Then, is a continuous martingale, and (3.20) holds.*

*Proof. *Lemma 3.16 asserts (3.19). The implications ((3.21) and (UI1) (3.22)); ((3.22) and (3.19) (3.25)), were established in the proof of Lemma 3.12.

#### 4. Sequences of Martingales with Asymptotically Conditionally Independent Increments

Lemma 4.1. *Let for each be an -valued square integrable martingale on with respect to a flow and a sub--algebra of . Suppose that conditions (UI1) and (RC) are fulfilled for ,
**
and there exists a nonrandom number such that for all **
Then,
*

*Proof. *Conditions (RC) , (4.1), and (4.2) entail, by Lemma 3.9, relative compactness of in .

Denote , , , ,
Condition (RC) implies that
In view of (4.1) . Then, by ItΓ΄'s formula
Hence, recalling that , we get

By the definition of and by condition (4.3),
Consequently,
and . The right-hand side of the last equality being less than , the sequence is u.i., and so is by Lemma 3.11 whose conditions (those not postulated) we have verified. This together with (4.9) and (4.3) implies uniform integrability of . Now, (4.8) and inequality (4.9) show that has this property, too.

By construction and Lemma 10.4 in [4], is a martingale. Then, it follows from (4.9) that , which together with (4.10) and (4.8) yields
So, it suffices to show that

Obviously, for any real and
Hence, from (4.5), (4.9), we get
Now, (4.12) ensues from (4.6), (4.3), (4.2), and stochastic boundedness of the sequence .

Lemma 4.2. *Let for each be an -valued starting from zero locally square integrable martingale with respect to some flow and a sub--algebra of . Suppose that condition (RC) is fulfilled for ;
**
for any ; there exists a nonrandom function such that
**
for all and . Then, for any , relation (4.4) holds.*

*Proof. *Let us denote, only in this proof, , , (so that and . The evident inequality
and condition (4.15) show us that for any positive and , the sequence is u.i.

By assumption, there exists a sequence of stopping times such that a.s. and for each . Then, for any , and , ,
Writing , we deduce from (4.17) and (4.15) uniform integrability of the sequence