9.7. THE LEBESGUE INTEGRAL, L1 209

and the triangle inequality holds, ∣∣∣∣∫ f dµ

∣∣∣∣≤ ∫ | f |dµ. (9.4)

Also, for every f ∈ L1 (Ω) it follows that for every ε > 0 there exists a simple function ssuch that |s| ≤ | f | and

∫| f − s|dµ < ε.

Proof: First consider the claim that the integral is linear. It was shown above that theintegral is linear on Re

(L1 (Ω)

). Then letting a+ ib,c+ id be scalars and f ,g functions in

L1 (Ω) ,

(a+ ib) f +(c+ id)g = (a+ ib)(Re f + i Im f )+(c+ id)(Reg+ i Img)

= cRe(g)−b Im( f )−d Im(g)+aRe( f )+ i(bRe( f )+ c Im(g)+a Im( f )+d Re(g))

It follows from the definition that∫(a+ ib) f +(c+ id)gdµ =

∫(cRe(g)−b Im( f )−d Im(g)+aRe( f ))dµ

+i∫

(bRe( f )+ c Im(g)+a Im( f )+d Re(g)) (9.5)

Also, from the definition,

(a+ ib)∫

f dµ +(c+ id)∫

gdµ = (a+ ib)(∫

Re f dµ + i∫

Im f dµ

)+(c+ id)

(∫Regdµ + i

∫Imgdµ

)which equals

= a∫

Re f dµ−b∫

Im f dµ + ib∫

Re f dµ + ia∫

Im f dµ

+c∫

Regdµ−d∫

Imgdµ + id∫

Regdµ−d∫

Imgdµ.

Using Lemma 9.7.5 and collecting terms, it follows that this reduces to 9.5. Thus theintegral is linear as claimed.

Consider the claim about approximation with a simple function. Letting h equal anyof

(Re f )+ ,(Re f )− ,(Im f )+ ,(Im f )− , (9.6)

It follows from the monotone convergence theorem and Theorem 8.1.6 on Page 181 thereexists a nonnegative simple function s ≤ h such that

∫|h− s|dµ < ε

4 . Therefore, lettings1,s2,s3,s4 be such simple functions, approximating respectively the functions listed in9.6, and s≡ s1− s2 + i(s3− s4) ,∫

| f − s|dµ ≤∫ ∣∣(Re f )+− s1

∣∣dµ +∫ ∣∣(Re f )−− s2

∣∣dµ

+∫ ∣∣(Im f )+− s3

∣∣dµ +∫ ∣∣(Im f )−− s4

∣∣dµ < ε

9.7. THE LEBESGUE INTEGRAL, L! 209and the triangle inequality holds,| / fay| < [ifldu. (9.4)Also, for every f € L' (Q) it follows that for every € > 0 there exists a simple function ssuch that |s| <|f| and f|f—s|du<e.Proof: First consider the claim that the integral is linear. It was shown above that theintegral is linear on Re (L! (Q)) . Then letting a+ ib,c+id be scalars and f, g functions in(a+ib) f +(c+id) g = (a+ib) (Re f+ilmf)+(c+id) (Reg +ilmg)= cRe(g) —bIm(f) —dIm(g) +aRe(f) +i(bRe(f) +cIm(g) +aIm(f) +dRe(g))It follows from the definition that[ (a+ ib) f+ (c+id) gd = | (cRe(g) ~bim(f) ~dtm(g) +aRe(f)) du+i [ (pRe(f) +elm(g) +alm(f) +dRe(g)) (9.5)Also, from the definition,(a+ ib) [ fau+(c+id) [ gd = (a+ib) ( [Re ran +i [ imfay)+(c+id) ( /Reeau +i [mean )which equals— a [Re fay —b [imfau +ib [Re fan +ia fm fa+e [ Regdu—d [ Imgdu+id [ Regd —d [ Imgdy.Using Lemma 9.7.5 and collecting terms, it follows that this reduces to 9.5. Thus theintegral is linear as claimed.Consider the claim about approximation with a simple function. Letting 4 equal anyof(Ref)* (Ref) ,(Imf)*,(imf)~, (9.6)It follows from the monotone convergence theorem and Theorem 8.1.6 on Page 181 thereexists a nonnegative simple function s < h such that f |h—s|du < §. Therefore, lettingS1,82,53,54 be such simple functions, approximating respectively the functions listed in9.6, and s = 5; — sy +i(s3—s4),Jif-siaws [ (Rep)! —siJau+ (Ref) —s2|au+ [\ampy*—ss|du [ (ims) ~s4|du <e