76.7. OTHER EXAMPLES, INCLUSIONS 2585

Therefore,∥u∥2

L2(U)+∫

U∑i, j

ai j (x)∂iu∂ ju = ∥u∥2W =

= ( f ,u)≤ ∥ f∥L2(U) ∥u∥L2(U) ≤ ∥ f∥L2(U) ∥u∥Wwhich shows that the map B−1 : H = L2 (U)→W is continuous.

Next suppose that Φ∈ L∞([0,T ]×Ω;L2

(Q1/2U,H

)). Then by continuity of the map-

ping B−1, it follows that Ψ ≡ B−1Φ satisfies Ψ ∈ L∞([0,T ]×Ω;L2

(Q1/2U,W

)). Thus

Φ = BΨ. In addition to this, to simplify the presentation, assume in addition that

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

V

⟨A(t,u,ω) ,u⟩ ≥ δ2 ∥u∥p

V

Also assume the uniqueness condition of Lemma 76.3.16 is satisfied. Consider the follow-ing graph.

−1/n

1/n

There is a monotone Lipschitz function Jnwhich is approximating a function with theindicated jump. For a convex function φ , we denote by ∂φ its subgradient. Thus fory ∈ ∂φ (x)

(y,u)≤ φ (x+u)−φ (x) .

Denote the Lipschitz function as Jn and the maximal monotone graph which it is approxi-mating as J. Thus J denotes the ordered pairs (x,y) which are of the form (0,y) for |y| ≤ 1along with ordered pairs (x,1) ,x > 0 and ordered pairs (x,−1) for x < 0. The graph of Jis illustrated in the above picture and is a maximal monotone graph. Thus J = ∂φ whereφ (r) = |r|. As illustrated in the graph, Jn is piecewise linear.

Let φ n (r)≡∫ r

0 Jn (s)ds. It follows easily that φ n (r)→ φ (r) uniformly on R. Also leth≥ 0 be progressively measurable and uniformly bounded by M and let u0 ∈ L2 (Ω,W ) ,u0F0 measurable. Then from the above theorems, there exists a unique solution to the integralequation

Bun (t)−Bu0 +∫ t

0A(s,un,ω)ds+

∫ t

0h(s,ω)Jn (un)ds = B

∫ t

0ΨdW,

the last term equaling∫ t

0 ΦdW. The integral equation holds off a set of measure zero and isprogressively measurable.

Then from the Ito formula, one obtains, using the monotonicity of Jn an estimate inwhich C does not depend on n

12

E ⟨Bun (t) ,un (t)⟩−12

E ⟨Bu0,u0⟩+E∫ t

0⟨Aun,un⟩V ds≤C

76.7. OTHER EXAMPLES, INCLUSIONS 2585Therefore,lela + ff Seago) Aude = luli =i,j= (fu) <|IFllieqy llellae cay S UF lln2 wy llellwwhich shows that the map B~! : H = L? (U) > W is continuous.Next suppose that ® € L® ([0,7] x Q; Z (Q'/?U,H)) . Then by continuity of the map-ping B~!, it follows that ¥ = B~'® satisfies ¥ € L® ([0,7] x Q;Z (Q'/2U,W)). Thus® = BY. In addition to this, to simplify the presentation, assume in addition that(A (t,u, @) —A (t,y, @) ,U—V) 2 o [ue —v|lp(A(t,u,@) ,u) > 5° |lu\lyAlso assume the uniqueness condition of Lemma 76.3.16 is satisfied. Consider the follow-ing graph.iT1/n=1/nThere is a monotone Lipschitz function J, which is approximating a function with theindicated jump. For a convex function @, we denote by 0@ its subgradient. Thus fory € Ag (x)(Vu) <O(x+u)—9 (x).Denote the Lipschitz function as J, and the maximal monotone graph which it is approxi-mating as J. Thus J denotes the ordered pairs (x,y) which are of the form (0,y) for |y| < 1along with ordered pairs (x, 1) ,x > 0 and ordered pairs (x,—1) for x < 0. The graph of Jis illustrated in the above picture and is a maximal monotone graph. Thus J = 0@ where (r) = |r|. As illustrated in the graph, J, is piecewise linear.Let ¢,,(r) = Jo Jn (s) ds. It follows easily that @, (r) + @(r) uniformly on R. Also leth > 0 be progressively measurable and uniformly bounded by M and let up € L” (Q,W) ,uoFo measurable. Then from the above theorems, there exists a unique solution to the integralequationt ot tBun (t) — Buy + | A(s,tm,0)d5-+ [ 1 (,0)Jn (tn) ds=B | Wdw,0 0 0the last term equaling fo @dW. The integral equation holds off a set of measure zero and isprogressively measurable.Then from the Ito formula, one obtains, using the monotonicity of J, an estimate inwhich C does not depend on n1 1 ‘t5F (Buy (t) ,Un (t)) — 56 (Buo, uo) +E | (Aun, Un)y ds <C