2160 CHAPTER 63. THE QUADRATIC VARIATION OF A MARTINGALE

Thus the expected value of the first sum equals 0. Similarly, the expected value of the sec-ond sum equals 0. Hence this has shown that for any bounded right continuous martingale,{ξ (s)} and t > 0,

E(ξ (t)

(A(t)−A′ (t)

))= 0.

Now let ξ be a bounded random variable and let ξ (t) be a right continuous version of themartingale E (ξ |Ft) . Then

0 = E(E (ξ |Ft)

(A(t)−A′ (t)

))= E

(E(ξ(A(t)−A′ (t)

)|Ft))

= E(ξ(A(t)−A′ (t)

))and since ξ is arbitrary, it follows that A(t) = A′ (t) a.e. which proves uniqueness.

Because of the uniqueness assertion, it suffices to prove the theorem on an arbitraryinterval, [0,a].

Without loss of generality, it can be assumed X (0) = 0 since otherwise, you could sim-ply consider X (t)−X (0) in its place and then at the end, add X (0) to M (t) . Let

{tnk

}mnk=0

be a sequence of partitions of [0,a] such that these are equally spaced points, limn→∞ tnk+1−

tnk = 0, and

{tnk

}mnk=0 ⊆

{tn+1k

}mn+1k=0 . Then consider the submartingale,

{X(tnk

)}mnk=0 . Theo-

rem 63.7.1 implies there exists a unique martingale, and increasing submartingale,

{M (tnk )}

mnk=0 and {A(tn

k )}mnk=0

respectively such that M (0) = 0 = A(0) ,

X (tnk ) = Mn (tn

k )+An (tnk ) .

and An(tnk

)is Ftn

k−1measurable. Recall how these were defined.

An (tnk ) =

k

∑j=1

E(

X(tn

j)−X

(tn

j−1)|Ftn

j−1

), An (0) = 0

Mn (tnk ) = X (tn

k )−An (tnk ) .

I want to show that {An (a)} is equi integrable. From this there will be a weakly conver-gent subsequence and nice things will happen. Define T n (ω) to equal tn

j−1 where tnj is the

first time where An(

tnj ,ω)≥ λ or T n (ω) = a if this never happens. I want to say that T n is

a stopping time and so I need to verify that[T n ≤ tn

j

]∈Ftn

jfor each j. If ω ∈

[T n ≤ tn

j

],

then this means the first time, tnk , where An

(tnk ,ω

)≥ λ is such that tn

k ≤ tnj+1. Since An

k isincreasing in k, [

T n ≤ tnj]

= ∪ j+1k=0 [A

n (tnk )≥ λ ]

=[An (tn

j+1)≥ λ

]∈Ftn

j.

Note T n only has the values tnk . Thus for t ∈ [tn

j−1, tnj ),

[T n ≤ t] =[T n ≤ tn

j−1]∈Ftn

j−1⊆Ft .

2160 CHAPTER 63. THE QUADRATIC VARIATION OF A MARTINGALEThus the expected value of the first sum equals 0. Similarly, the expected value of the sec-ond sum equals 0. Hence this has shown that for any bounded right continuous martingale,{& (s)} andr > 0,E (E(t) (A() A") =0.Now let € be a bounded random variable and let & (+) be a right continuous version of themartingale E (&|.F,). ThenE(E(§|F,) (A(t)—A'(1))) =E(E(E (A() —4’(1)) |F))E(§(A(t)—A’(t)))and since € is arbitrary, it follows that A(t) = A’ (t) a.e. which proves uniqueness.Because of the uniqueness assertion, it suffices to prove the theorem on an arbitraryinterval, [0, a].Without loss of generality, it can be assumed X (0) = 0 since otherwise, you could sim-ply consider X (t) — X (0) in its place and then at the end, add X (0) to M(t). Let {27 cobe a sequence of partitions of [0,a] such that these are equally spaced points, limy_,.. ¢/! “aTt; = 0, and {tr ye C {rl ye . Then consider the submartingale, {X (t?) 10° Theo-rem 63.7.1 implies there exists a unique martingale, and increasing submartingale,0{M (th) eto and {A (2) }%orespectively such that M (0) =0 =A(0),X (i) = M" (i) +A" (tq).and A” (1?) is Fin measurable. Recall how these were defined.kAN) = VE (x (##) =x (11) |Fn_,). A") =0J=M" (te) =X (tq) —A" (th) -I want to show that {A” (a) } is equi integrable. From this there will be a weakly conver-gent subsequence and nice things will happen. Define T” (@) to equal thy where fis thefirst time where A” («, o) >A or T" (@) =a if this never happens. I want to say that T” isa stopping time and so I need to verify that ia < "| € Fy for each j. If @ € a < "| ,then this means the first time, ¢?, where A” (tf, ) > A is such that es t} ‘yy. Since Aj isincreasing in k,[T’<eh] = UA") =A]= [A (ih) >A] e Hy.Note T” only has the values t. Thus for t € [- tt),[T’ <i] =[T" <r Je Fy, oF,