2296 CHAPTER 67. THE EASY ITO FORMULA

Claim: limn→∞ X[0,τn] = 1.Proof of claim: From maximal estimates as in the construction of the stochastic integral

and the Borel Cantelli lemma, it follows that there exists a subsequence still denoted by nand a set of measure zero N such that for ω /∈ N1,∫ t

0ΦndW →

∫ t

0ΦdW

uniformly on [0,T ] . Also one can show that off a set of measure zero, there is a subsequencestill called n such that

∫ t0 φ n (s)ds→

∫ t0 φ (s)ds uniformly on [0,T ] . Here is why.

E(∣∣∣∣∫ t

0φ n (s)ds−

∫ t

0φ (s)ds

∣∣∣∣)≤ ∫Ω

∫ T

0|φ n−φ |dtdP

which is given to converge to 0. Thus

P(

maxt∈[0,T ]

∣∣∣∣∫ t

0φ n (s)ds−

∫ t

0φ (s)ds

∣∣∣∣> λ

)≤ P

(∫ T

0|φ n (s)−φ (s)|ds > λ

)

≤ 1λ

∫[∫ T

0 |φn(s)−φ(s)|ds>λ ]

∫ T

0|φ n (s)−φ (s)|dsdP

≤ 1λ

∫Ω

∫ T

0|φ n (s)−φ (s)|dsdP

Thus

P(

maxt∈[0,T ]

∣∣∣∣∫ t

0φ n (s)ds−

∫ t

0φ (s)ds

∣∣∣∣> 2−k)≤ 2k

∫Ω

∫ T

0|φ n (s)−φ (s)|dsdP

If n > nk, the right side is less than 2−k. Use φ nk. Then there exists a set of measure zero

N2 such that for ω /∈ N2, ∣∣∣∣∫ t

0φ n (s)ds−

∫ t

0φ (s)ds

∣∣∣∣→ 0

uniformly. Hence, you can take a couple of subsequences and assert that there exists asubsequence still called n and a set of measure zero N such that Xn (t)→ X (t) uniformlyon [0,T ] for each ω /∈ N. Since |X (t,ω)| < M, it follows that for each ω /∈ N, when n islarge enough, τn = ∞ and this proves the claim.

From the claim, it follows that X[0,τn]Φn→ Φ◦ J−1 in L2([0,T ]×Ω;L2

(Q1/2U,H

))and X[0,τn]φ n → φ in L1 ([0,T ]×Ω;H ) . Thus you can replace Φn in the above withX[0,τn]Φn and φ n with X[0,τn]φ n. Thus there exists a subsequence, still called n and aset of measure zero N such that for ω /∈ N,∫ t

0X[0,τn]ΦndW →

∫ t

0ΦdW

2296 CHAPTER 67. THE EASY ITO FORMULAClaim: lim, ,.. %/o,2,] = 1.Proof of claim: From maximal estimates as in the construction of the stochastic integraland the Borel Cantelli lemma, it follows that there exists a subsequence still denoted by nand a set of measure zero N such that for @ ¢ Ni,t ti ow — | DddwJO JOuniformly on [0,7]. Also one can show that off a set of measure zero, there is a subsequencestill called n such that {5 @,, (s)ds + J} ¢ (s)ds uniformly on [0,7]. Here is why.e(|[ eta [oval) < f° lon —olaarwhich is given to converge to 0. ThusP ( max [oes )as— [ g (s) ds >a) <P([ ld, (s )=9(s)\as>A)< a Thos. J |On(8)-9(s)|ds>A] ah Pn (s)|dsdP< iff lenls)-0(asarThusP ( max [ , ( sjds— ['9() )ds| >27 ‘Jax 10, (s) 6 (s)|dsdPIf n > ng, the right side is less than 2~*. Use @,,- Then there exists a set of measure zeroNy» such that for @ ¢ Np,t t[[eairas~ ['o()as0 0uniformly. Hence, you can take a couple of subsequences and assert that there exists asubsequence still called n and a set of measure zero N such that X, (t) — X (t) uniformlyon [0,7] for each ¢ N. Since |X (t,@)| < M, it follows that for each w ¢ N, when n islarge enough, T,, = ° and this proves the claim.From the claim, it follows that 279 7,)Pn + oJ”! in L? ((0,T] x Q:-4 (Q'/?U,H))and 29.7,)0, 4 @ in L! ((0,7] x Q;). Thus you can replace ®, in the above with210,t,JPn and >, with 2o.7,)9,- Thus there exists a subsequence, still called n and aset of measure zero N such that for @ ¢ N,>0ot ot| ZocaPraw + | eaw0 : 0