76.7. OTHER EXAMPLES, INCLUSIONS 2593

where Φ ∈ L∞((0,T )×Ω,L2

(Q1/2U,H

))for H = L2 (0,1). As in the previous example,

simple considerations involving maximal monotone operators imply that there exists Ψ ∈L∞((0,T )×Ω,L2

(Q1/2U,W

))such that Φ = BΨ. We also assume that f ∈ V ′ and u0,v0

are in L2 (Ω,V ) and L2 (Ω,W ) respectively, both being F0 measurable. The above equationdoes not fit the general theory developed earlier because it is second order in time and isa stochastic version of a hyperbolic equation rather than a parabolic one. We consider itusing a parabolic regularization which can be studied with the above general theory alongwith a simple fixed point argument.

Consider the approximate problem which is to find a solution to

Bv(t)−Bv0 + ε

∫ t

0Avds+

∫ t

0A(u)ds =

∫ t

0f ds+

∫ t

0ΦdW (76.7.71)

where u is given above as an integral of v. First we argue that there exists a unique solutionto the above integral equation and then we pass to a limit as ε → 0. Let u ∈ V be given.

From Corollary 76.4.9 there exists a unique solution v to 76.7.71. Now suppose u1,u2are two given in V and denote by vi the corresponding v which solves the above. Thenfrom the Ito formula or standard considerations,

12

E ⟨B(v1 (t)− v2 (t)) ,v1 (t)− v2 (t)⟩+ εE∫ t

0∥v1− v2∥2

V ds

≤ ε

2E∫ t

0∥v1− v2∥2

V ds+Cε E∫ t

0∥u1−u2∥2

V ds

Now define a mapping θ from V to V as follows. Begin with v then obtain

u(t)≡ u0 +∫ t

0v(s)ds (76.7.72)

Use this u in 76.7.71. Then θv is the solution to 76.7.71 which corresponds to u. Then theabove inequality shows that∫ t

0

∫Ω

∥θv1 (s)−θv2 (s)∥2 dPds≤ Cε

ε

∫ t

0

∫Ω

∥u1−u2∥2V dPds

≤ Cε

εCT

∫ t

0

∫ s

0

∫Ω

∥v1 (r)− v2 (r)∥2V dPdrds

It follows that a high enough power of θ is a contraction map on L2(0,T,L2 (Ω,V )

)and

so there exists a unique fixed point. This yields a unique solution to the above approximateproblem 76.7.71 in which u,v are related by 76.7.72.

Next we let ε → 0. Index the above solution with ε . By the Ito formula again,

12

E ⟨Bvε (t) ,vε (t)⟩−12

E ⟨Bv0,v0⟩+ ε

∫ t

0E ∥vε∥2

V ds

+12

E ∥uε (t)∥2V −

12

E ∥u0∥2V =

∫ t

0E ⟨ f ,vε⟩ds

76.7. OTHER EXAMPLES, INCLUSIONS 2593where ® € L® ((0,T) x Q,.Z (Q'/?U,H)) for H = L? (0,1). As in the previous example,simple considerations involving maximal monotone operators imply that there exists ¥ €Ll” (0, T)xQ,4 (Q'/?U,W)) such that ® = BY. We also assume that f € WV’ and uo, voare in L? (Q,V) and L? (Q,W) respectively, both being .7p measurable. The above equationdoes not fit the general theory developed earlier because it is second order in time and isa stochastic version of a hyperbolic equation rather than a parabolic one. We consider itusing a parabolic regularization which can be studied with the above general theory alongwith a simple fixed point argument.Consider the approximate problem which is to find a solution tot t t tBr(t)—Bup +e [ Avds+ | A(u)ds= [ fas+ | bdW (76.7.71)0 0 0 0where u is given above as an integral of v. First we argue that there exists a unique solutionto the above integral equation and then we pass to a limit as € > 0. Let u € V be given.From Corollary 76.4.9 there exists a unique solution v to 76.7.71. Now suppose u1,u2are two given in Y and denote by v; the corresponding v which solves the above. Thenfrom the Ito formula or standard considerations,1 t5E(B(v1 ¢) —va(t)) v1 (v2 (0) +e f I —vallpas€, f' 2 ’ 2< SE f Im —vollpds +CcE [lus —uallyds0 0Now define a mapping @ from ¥ to ¥ as follows. Begin with v then obtainu(t) =uo+ [ v(s)ds (76.7.72)Use this uv in 76.7.71. Then @v is the solution to 76.7.71 which corresponds to u. Then theabove inequality shows thatt t[Lien (s) ~ v2 (s)[?aPds << | | lie) —up||2 dPas0 JQ E Jo JQCe ef rs ao 2< cr | | / Iv (7) — v2 (V)II2 dPdrdsE 0 Jo JaIt follows that a high enough power of @ is a contraction map on L? (0, T,L? (Q,V)) andso there exists a unique fixed point. This yields a unique solution to the above approximateproblem 76.7.71 in which u,v are related by 76.7.72.Next we let € — 0. Index the above solution with €. By the Ito formula again,1 1 t5 (Bre (0) .¥e (0) ~ 5E (Bvo.v0) +e ff E lvelly ds1 1 ‘t+3 ue (|i — 5 lluolly = [) (Fave) as