76.7. OTHER EXAMPLES, INCLUSIONS 2591

Theorem 76.7.2 In the situation of Corollary 76.4.9 where V is a closed subspace ofW σ ,p (U) ,σ > 1 and W is as described above for U a bounded open set, u0 ∈ L2 (Ω,W ) , u0F0 measurable. Suppose λ I +A(t,u,ω) satisfies

⟨λ I +A(t,u,ω)− (λ I +A(t,v,ω)) ,u− v⟩ ≥ δ2 ∥u− v∥p

V

for all λ large enough. Also assume Φ ∈ L∞([0,T ]×Ω,L2

(Q1/2U,H

))with Φ = BΨ

whereΨ ∈ L∞

([0,T ]×Ω,L2

(Q1/2U,W

))and progressively measurable. Then there exists a unique solution to the integral equation

Bu(t)−Bu0 +∫ t

0A(s,u)ds+

∫ t

0h(s,ω)ζ (s,ω)ds =

∫ t

0ΦdW (76.7.70)

where for a.e. s, h(s,ω)ζ (s,ω) ∈ ∂u (h(s,ω)φ (u(s))) where φ (r)≡ |r|. The symbol ∂u isthe subgradient of φ (u). Written in terms of inclusions, there exists a set of measure zerosuch that off this set,(

B(

u−∫ (·)

0ΦdW

))′+A(t,u) ∈ ∂u (h(t,ω)φ (u(s))) a.e. t

u(0) = u0

Note that one can replace

Φ ∈ L∞

([0,T ]×Ω,L2

(Q1/2U,H

))with Φ ∈ L2

([0,T ]×Ω,L2

(Q1/2U,H

))along with an assumption that t → Φ(t,ω) is

continuous. This can be done by defining a stopping time

τn ≡ inf{t : ∥Φ(t)∥> n}

Then from the above example, there exists a solution to the integral equation off a set ofmeasure zero

Bun (t)−Bu0 +∫ t

0A(s,un)ds+

∫ t

0h(s,ω)ζ n (s,ω)ds =

∫ t∧τn

0ΦdW

Since Φ is a continuous process, τn = ∞ for all n large enough. Hence, one can replace theabove with the desired integral equation. Of course the size of n depends on ω, but we candefine

u(t,ω) = limn→∞

un (t,ω)

because by uniqueness which comes from monotonicity, if for a particular ω, both n,kare sufficiently large, then un = uk. Thus u is progressively measurable and is the desiredsolution.

Next we show that the above theory can also be used as a starting point for some secondorder in time problems. Consider a beam which has a point mass of mass m attached to

76.7. OTHER EXAMPLES, INCLUSIONS 2591Theorem 76.7.2 In the situation of Corollary 76.4.9 where V is a closed subspace ofW°? (U),o > LandW is as described above for U a bounded open set, ug € L? (Q,W) , uoFo measurable. Suppose AI +A (t,u, @) satisfies(AI+A (t,u,@) —(AI+A(t,v,@)) ,u—v) > 8? \lu— vl?for all 2 large enough. Also assume ® € L” ({0, T|xXQ,L4 (Q'/?U,H)) with ® = BYwhereWel ((o. T]xQ,P, (o'u,w))and progressively measurable. Then there exists a unique solution to the integral equationt t tBu (t) — Bug + | A(s,u)ds-+ i h(s,@) (s,@) ds = | dW (76.7.70)0 0 0where for a.e. s,h(s,@) ¢ (s,@) € 0, (h(s, @) d (u(s))) where 9 (r) = |r|. The symbol 0, isthe subgradient of @(u). Written in terms of inclusions, there exists a set of measure zerosuch that off this set,(8 (« - [ aw) +A(tu) € a(h(t,@)o(u(s))) ae tu(0) = uoNote that one can replace@EL” ((0.71 x2, DP (0'u,H))with ® € L? ((0,T] x Q,% (Q'/?U,H)) along with an assumption that t > ®(1,@) iscontinuous. This can be done by defining a stopping timeT, =inf{t:||®(1)|| >n}Then from the above example, there exists a solution to the integral equation off a set ofmeasure zerotTt tBun (t) — Buy + [| A(s,un)ds-+ f h(s,@)C,,(s,@)ds = @dw0 0 0Since ® is a continuous process, T, = °° for all m large enough. Hence, one can replace theabove with the desired integral equation. Of course the size of n depends on @, but we candefineu(t, @) = lim un (t,@)because by uniqueness which comes from monotonicity, if for a particular @, both n,kare sufficiently large, then u, = uz. Thus uw is progressively measurable and is the desiredsolution.Next we show that the above theory can also be used as a starting point for some secondorder in time problems. Consider a beam which has a point mass of mass m attached to