62.7. DOOB OPTIONAL SAMPLING CONTINUOUS CASE 2087

To verify that |X (τ)| is integrable, note that by right continuity, X (τn)→ X (τ) point-wise. Apply the Vitali convergence theorem to obtain∫

|X (τ)|dP = limn→∞

∫Ω

|X (τn)|dP≤∫

X (T )dP < ∞.

In fact, you do not need to assume X is nonnegative.

Lemma 62.7.13 Let X (t) be a right continuous submartingale such that the filtration {Ft}is normal. Recall this includes

Ft = ∩s>tFs.

Also let τ be a stopping time with values in [0,T ] . Let Pn ={

tnk

}mn+1k=1 be a sequence of

partitions of [0,T ] which have the property that

Pn ⊆Pn+1, limn→∞||Pn||= 0,

where||Pn|| ≡ sup

{∣∣tnk − tn

k+1

∣∣ : k = 1,2, · · · ,mn}

Then let

τn (ω)≡mn

∑k=0

tnk+1Xτ−1((tn

k ,tnk+1])

(ω)

It follows that τn is a stopping time and also the functions |X (τn)| are uniformly integrable.Furthermore, |X (τ)| is integrable.

Proof: It was shown above that τn is a stopping time. Also, tnk → X

(tnk

)is a discrete

submartingale. Then by Theorem 62.6.6 there is a martingale tnk →M

(tnk

)and an increasing

submartingale tnk → A

(tnk

)such that A≥ 0 and is increasing

X (tnk ) = M (tn

k )+A(tnk )

You define A(tm0)≡ 0 and for n≥ 1,

A(tmn )≡

n

∑k=1

E(

X (tmk )−X

(tmk−1)|Ftm

k−1

)and repeat the arguments in that theorem. You know that A(0) ,A(τn) ,A(T ) is a submartin-gale by the optional sampling theorem given earlier, Theorem 62.6.7, and so

P(A(τn)> λ )≤ 1λ

∫[A(τn)>λ ]

A(τn)dP≤ 1λ

∫[A(τn)>λ ]

A(T )dP≤ ∥A(T )∥L1

λ

It also follows from the definition of A that

∥A(T )∥L1 =∫

X (T )−X (0)dP < ∞

Hencelim

λ→∞

∫[A(τn)>λ ]

A(τn)dP≤ limλ→∞

∫[A(τn)>λ ]

A(T )dP = 0

62.7. DOOB OPTIONAL SAMPLING CONTINUOUS CASE 2087To verify that |X (t)| is integrable, note that by right continuity, X (t,) — X (Tt) point-wise. Apply the Vitali convergence theorem to obtain[ |X (1) |dP = lim | IX (t,)|dP < i X(T)dP<o. fjJQ nro JO, JQIn fact, you do not need to assume X is nonnegative.Lemma 62.7.13 Let X (t) be a right continuous submartingale such that the filtration { F,}is normal. Recall this includest= Asst Fs.Also let t be a stopping time with values in [0,T]. Let Py = {i mt be a sequence ofpartitions of |0,T| which have the property thatPn SC Ani, lim || F,|| =,n—-oowhereI| Pal] = sup {| — tyr] sk = 1,2,--+ ,m0Then letMpT(@) = Yt Fc !) (@)k=0nN ynFotiIt follows that Ty, is a stopping time and also the functions |X (T,)| are uniformly integrable.Furthermore, |X (t)| is integrable.Proof: It was shown above that T, is a stopping time. Also, t/ > X (tf) is a discretesubmartingale. Then by Theorem 62.6.6 there is a martingale t + M (t”) and an increasingsubmartingale t/’ — A (t/) such that A > 0 and is increasingX (th) =M (i) +A (te)You define A (¢7”) = 0 and for n > 1,A(t) = E(x) —X (4) Fe)k=1and repeat the arguments in that theorem. You know that A (0) ,A (t,,) ,A (7) is a submartin-gale by the optional sampling theorem given earlier, Theorem 62.6.7, and so1 1 A(T) [lrP(A(T)>A)<~ A(m)dP <> | A(T)dp< “Ve(A(tu) > A) A J[A(tn)>A] (Fn) [A(tn) >A] 7) AIt also follows from the definition of A that|A(T)Il = [.X(1)—X O)aP <~Hencelim A(t,)dP < lim A(T)dP =0A900 J[A( tp) >A] A900 A(t) >A]