698 CHAPTER 24. THE BOCHNER INTEGRAL

Proof: It suffices to show S has an η net in Lp ([a,b] ;W ) for each η > 0.If not, there exists η > 0 and a sequence {un} ⊆ S, such that

∥un−um∥ ≥ η (24.49)

for all n ̸= m and the norm refers to Lp ([a,b] ;W ). Let

a = t0 < t1 < · · ·< tk = b, ti− ti−1 = (b−a)/k.

Now define un (t)≡∑ki=1 uniX[ti−1,ti) (t) , uni ≡ 1

ti−ti−1

∫ titi−1

un (s)ds. The idea is to show thatun approximates un well and then to argue that a subsequence of the {un} is a Cauchysequence yielding a contradiction to the above ∥un−um∥ ≥ η .

Therefore,

un (t)−un (t) =k

∑i=1

un (t)X[ti−1,ti) (t)−k

∑i=1

uniX[ti−1,ti) (t)

=k

∑i=1

1ti− ti−1

∫ ti

ti−1

un (t)dsX[ti−1,ti) (t)−k

∑i=1

1ti− ti−1

∫ ti

ti−1

un (s)dsX[ti−1,ti) (t)

=k

∑i=1

1ti− ti−1

∫ ti

ti−1

(un (t)−un (s))dsX[ti−1,ti) (t) .

It follows from Jensen’s inequality, Lemma 10.15.1 that

||un (t)−un (t)||pW =k

∑i=1

∥∥∥∥ 1ti− ti−1

∫ ti

ti−1

(un (t)−un (s))ds∥∥∥∥p

WX[ti−1,ti) (t)

≤k

∑i=1

1ti− ti−1

∫ ti

ti−1

∥un (t)−un (s)∥pW dsX[ti−1,ti) (t)

and so ∫ b

a∥un (t)−un (s)∥p

W ds

≤∫ b

a

k

∑i=1

1ti− ti−1

∫ ti

ti−1

∥un (t)−un (s)∥pW dsX[ti−1,ti) (t)dt

=k

∑i=1

1ti− ti−1

∫ ti

ti−1

∫ ti

ti−1

∥un (t)−un (s)∥pW dsdt. (24.50)

From Lemma 24.11.1 if ε > 0, there exists Cε such that

∥un (t)−un (s)∥pW ≤ ε ∥un (t)−un (s)∥p

E +Cε ∥un (t)−un (s)∥pX

≤ 2p−1ε (∥un (t)∥p +∥un (s)∥p)+Cε |t− s|p/q

This is substituted in to 24.50 to obtain∫ b

a∥un (t)−un (s)∥p

W ds≤

698 CHAPTER 24. THE BOCHNER INTEGRALProof: It suffices to show S has an 77 net in L? ([a,b];W) for each n > 0.If not, there exists 7 > 0 and a sequence {u,,} C S, such thatI]un — Um|| = 7 (24.49)for all n 4 m and the norm refers to L? ([a,b] ;W). Leta=t9 <t) <-++ <%=), t;-t-1 = (b—a) /k.Now define a (t) = Yi) Wn, Kir, yj) (t)> Un, = mo fe _ Un (s) ds. The idea is to show thatU, approximates u,, well and then to argue that a subsequence of the {Z,} is a Cauchysequence yielding a contradiction to the above ||u, —um|| > 1.Therefore,k kUn (t) — Un (t) = Ye tn (t) Bj, [t;—1,t7) )— Ltn By ti)i=l i=l1 tj k tj7 ares [ tin (1) ds, 1) -)h ti—tj-1 col ' n(s)ds2iy,_, 4) (t)i-1 i=lk tj—— f° (unt) —tn(s)) ds By. 44) (0)1 fi ti 1 Jtj_-1It follows from Jensen’s inequality, Lemma 10.15.1 that_ k 1 ti Pln (0) Fin (Me = YJ —— (en) tn (9) 5] Fievei=l fi i—1 JYtj—-] WwWSL bom lteastin vayi= ‘i i-1 Jtj-]and sob; \|Un (t) — tin (8) lly dsbk 1 tj >< | bi = J lie I dB gay (Oea i=l“ Llk tj=) — | [ ||un (t) — Un (s)||h, dsdt. (24.50)ja fi — fi-1 ti) Ytj-1From Lemma 24.11.1 if € > 0, there exists Cg such thatIlun (t) — Un (8) lly S € |In (1) = Un (8) [lz + Ce ||Un (4) = Un (8) Ik<2? HE (lun (1)? + lun (8) |?) + Ce |e — |?"This is substituted in to 24.50 to obtain[into — Un (s (s)||y ds <