陶哲轩实分析(上)11.8及习题-Analysis I 11.8

Exercise 11.8.1

We prove this by induction on n n . More precisely, we let P ( n ) P(n) be the property that whenever I I is a bounded interval, and whenever P \mathbf P is a partition of I I with cardinality n n , that α [ I ] = J P α [ J ] α[I]=\sum_{J\in \mathbf P}α[J] .
The case P ( 0 ) P(0) and P ( 1 ) P(1) are all easy to prove. Now suppose inductively that P ( n ) P(n) is true for n 1 n\geq 1 , let I I be a bounded interval, and P \mathbf P be a partition of I I with cardinality n + 1 n+1 .
If I I is the empty set or a point, then all the intervals in P \mathbf P must also be either the empty set or a point and so every interval has length zero and the claim is trivial. Thus we will assume that I I is an interval of the form ( a , b ] , [ a , b ) , ( a , b ) (a,b],[a,b),(a,b) or [ a , b ] [a,b] .
Let us first suppose that b I b\in I , i.e., I I is either ( a , b ] (a,b] or [ a , b ] [a,b] . Since b I b\in I , we know that one of the intervals K K in P \mathbf P contains b b . Since K K is contained in I I , it must therefore be of the form ( c , b ] , [ c , b ] (c,b],[c,b] , or { b } \{b\} for some real number c c , with a c b a\leq c\leq b (in the latter case of K = { b } K=\{b\} , we set c : = b c:=b ). In particular, this means that the set I K I-K is also an interval of the form [ a , c ] , ( a , c ) , ( a , c ] , [ a , c ) [a,c],(a,c),(a,c],[a,c) when c > a c>a , or a point or empty set when a = c a=c . Either way, we easily see that
α [ I ] = α ( b ) α ( a ) = α ( b ) α ( c ) + α ( c ) α ( a ) = α [ K ] + α [ I K ] α[I]=α(b)-α(a)=α(b)-α(c)+α(c)-α(a)=α[K]+α[I-K]
On the other hand, since P \mathbf P forms a partition of I I , we see that P { K } \mathbf P-\{K\} forms a partition of I K I-K , By the induction hypothesis, we thus have
α [ I K ] = J P { K } α [ J ] α[I-K]=\sum_{J\in \mathbf P-\{K\}}α[J]
Combining these two identities, we obtain
α [ I ] = J P α [ J ] α[I]=\sum_{J\in \mathbf P}α[J]
Now suppose that b I b\notin I , i.e., I is either ( a , b ) (a,b) or [ a , b ) [a,b) , Then one of the intervals K K also is of the form ( c , b ) (c,b) or [ c , b ) [c,b) . In particular, this means that the set I K I-K is also an interval of the form [ a , c ] , ( a , c ) , ( a , c ] , [ a , c ) [a,c],(a,c),(a,c],[a,c) when c > a c>a , or a point or empty set when a = c a=c . The rest of the argument then proceeds as above.

Exercise 11.8.2

Proposition: Let I I be a bounded interval, and let f : I R f:I\to\mathbf R be a function. Suppose that P \mathbf P and P \mathbf P' are partitions of I I such that f f is piecewise constant both with respect to P \mathbf P and with respect to P \mathbf P' . Then p . c . [ P ] f d α = p . c . [ P ] f d α p.c.\int_{[\mathbf P]} fdα=p.c.\int_{[\mathbf P']}fdα .
Proof: By Lemma 11.2.7, we know f f is piecewise constant with respect to P # P \mathbf P\#\mathbf P' , thus the value
p . c . [ P # P ] f d α = J P # P c J α [ J ] p.c.\int_{[\mathbf P\#\mathbf P']}fdα=\sum_{J\in \mathbf P\# \mathbf P'}c_J α[J]
is well defined. Now choose any K P K\in \mathbf P , then P K = { J P # P : J K } \mathbf P_K=\{J\in \mathbf P\# \mathbf P':J\subseteq K\} is a partition of K K , and f f is constant with constant value c K c_K on both K K and all elements of P K \mathbf P_K , thus by Theorem 11.1.13 we have c J = c K , J P K c_J=c_K,\forall J\in \mathbf P_K , and
α [ K ] = J P K α [ J ]       c K α [ K ] = J P K c K α [ J ] = J P K c J α [ J ] α[K]=\sum_{J\in \mathbf P_K}α[J] \implies c_K α[K]=\sum_{J\in \mathbf P_K}c_K α[J]=\sum_{J\in \mathbf P_K}c_J α[J]
Also, consider the set { J P # P : J K  for some  K P } P # P \{J\in \mathbf P\# \mathbf P':J\subseteq K \text{ for some }K\in \mathbf P\}\subseteq \mathbf P\# \mathbf P' , for any J P # P J\in \mathbf P\# \mathbf P' , by definition we can find a K P K\in \mathbf P and a K P K'∈P' s.t. J = K K J=K\cap K' , so the two sets are equal. Thus
p . c . [ P ] f d α = K P c K α [ K ] = K P J P K c J α [ J ] = J P # P c J α [ J ] = p . c . [ P # P ] f d α \begin{aligned}p.c.\int_{[\mathbf P]}fdα&=\sum_{K\in \mathbf P}c_K α[K]=\sum_{K\in \mathbf P}\sum_{J\in P_K}c_J α[J]=\sum_{J\in \mathbf P\# \mathbf P'}c_J α[J]\\&=p.c.\int _{[\mathbf P\# \mathbf P']}fdα\end{aligned}
Similarly we can prove p . c . [ P ] f d α = p . c . [ P # P ] f d α p.c.\int_{[\mathbf P' ]}fdα=p.c.\int_{[\mathbf P\# \mathbf P' ]} fdα , and the statement is proved.

Exercise 11.8.3

Proposition: Let I I be a bounded interval, and let f : I R f:I→\mathbf R and g : I R g:I→\mathbf R be piecewise constant functions on I I . Let α : X R α:X→\mathbf R be defined on a domain containing I I which is monotone increasing, then
( a ) We have p . c . I ( f + g ) d α = p . c . I f d α + p . c . I g d α p.c.∫_I (f+g)dα=p.c.∫_I fdα+p.c.∫_I gdα .
( b ) For any real number c c , we have p . c . I ( c f ) d α = c ( p . c . I f d α ) p.c.∫_I (cf)dα=c(p.c.∫_I fdα) .
( c ) We have p . c . I ( f g ) d α = p . c . I f d α p . c . I g d α p.c.∫_I (f-g)dα=p.c.∫_I fdα-p.c.∫_I gdα .
( d ) If f ( x ) 0 , x I f(x)\geq 0,\forall x\in I , then p . c . I f d α 0 p.c.∫_I fdα\geq 0 .
( e ) If f ( x ) g ( x ) , x I f(x)\geq g(x),\forall x\in I , then p . c . I f d α p . c . I g d α p.c.∫_I fdα\geq p.c.∫_I gdα .
( f ) If f ( x ) = c , x I f(x)=c,\forall x\in I , then p . c . I f d α = c α [ I ] p.c.∫_I fdα=cα[I] .
( g ) Let J J be a bounded interval containing I I (i.e. I J I\subseteq J ), and let F : J R F:J→\mathbf R be the function
F ( x ) = { f ( x ) , x I 0 , x I F(x)=\begin{cases}f(x),&x\in I\\0,&x\notin I\end{cases}
Then F F is piecewise constant on J J , and p . c . J F d α = p . c . I f d α p.c.∫_J Fdα=p.c.∫_I fdα .
( h ) Suppose that { J , K } \{J,K\} is a partition of I I into two intervals J J and K K . Then the functions f J : J R f|_J:J→\mathbf R and f K : K R f|_K:K→\mathbf R are piecewise constant on J J and K K respectively, and we have
p . c . I f d α = p . c . J f J d α + p . c . K f K d α p.c.∫_I fdα=p.c.∫_Jf|_J dα+p.c.∫_Kf|_K dα
Proof:
We choose partitions of I : P I: \mathbf P' and P \mathbf P'' such that f f is piecewise constant with respect to P \mathbf P' and g g is piecewise constant with respect to P \mathbf P'' . Then let P = P # P \mathbf P=\mathbf P'\# \mathbf P'' , we can see f f and g g are piecewise constant with respect to P \mathbf P . For any K P K\in \mathbf P , let c K , d K c_K,d_K denote the constant value of f f and g g on K K .

( a ) f + g f+g is piecewise constant with respect to P \mathbf P , with constant value c K + d K c_K+d_K on K P K\in \mathbf P ,
p . c . I ( f + g ) d α = p . c . [ P ] ( f + g ) d α = K P ( c K + d K ) α [ K ] = K P c K α [ K ] + K P d K α [ K ] = p . c . [ P ] f d α + p . c . [ P ] g d α = p . c . I f d α + p . c . I g d α \begin{aligned}p.c.∫_I (f+g)dα&=p.c.∫_{[\mathbf P]} (f+g)dα=\sum_{K\in \mathbf P}(c_K+d_K)α[K]\\&=\sum_{K\in \mathbf P}c_K α[K]+\sum_{K\in \mathbf P}d_K α[K]\\&=p.c.∫_{[\mathbf P]} fdα+p.c.∫_{[\mathbf P]} gdα\\&=p.c.∫_I fdα+p.c.∫_I gdα\end{aligned}
( b ) c f cf is piecewise constant with respect to P \mathbf P , with constant value c c K cc_K on K P K∈\mathbf P ,
p . c . I ( c f ) d α = p . c . [ P ] ( c f ) d α = K P ( c c K ) α [ K ] = c K P c K α [ K ] = c ( p . c . [ P ] f d α ) = c ( p . c . I f d α ) \begin{aligned}p.c.∫_I (cf)dα&=p.c.∫_{[\mathbf P]} (cf)dα=∑_{K∈\mathbf P}(cc_K)α[K]\\&=c∑_{K∈\mathbf P}c_K α[K]=c\left(p.c.∫_{[\mathbf P]} fdα\right)\\&=c\left(p.c.∫_I fdα\right)\end{aligned}
( c ) Use ( b ) we have p . c . I ( g ) d α = p . c . I g d α p.c.∫_I (-g)dα=-p.c.∫_I gdα , then use ( a ) we get
p . c . I ( f g ) d α = p . c . I ( f + ( g ) ) d α = p . c . I f d α + p . c . I ( g ) d α = p . c . I f d α p . c . I g d α \begin{aligned}p.c.∫_I(f-g)dα&=p.c.∫_I (f+(-g))dα\\&=p.c.∫_I fdα+p.c.∫_I (-g)dα\\&=p.c.∫_Ifdα-p.c.∫_Igdα\end{aligned}
( d ) If f ( x ) 0 , x I f(x)≥0,∀x∈I , then c K 0 , K P c_K≥0,∀K∈\mathbf P , since α α is monotone increasing we have c K α [ K ] 0 , K P c_K α[K]≥0,∀K∈\mathbf P , thus
p . c . I f d α = p . c . [ P ] f d α = K P c K α [ K ] 0 p.c.∫_I fdα=p.c.∫_{[\mathbf P]} fdα=∑_{K∈P}c_K α[K]≥0
( e ) We have f ( x ) g ( x ) 0 , x I f(x)-g(x)≥0,∀x∈I , so use ( d ) we have
p . c . I ( f g ) d α = p . c . I f d α p . c . I g d α 0 p . c . I f d α p . c . I g d α p.c.∫_I (f-g)dα=p.c.∫_I fdα-p.c.∫_I gdα≥0 \\⇒ p.c.∫_I fdα≥p.c.∫_I gdα
( f ) If f ( x ) = c , x I f(x)=c,∀x∈I , then c K = c , K P c_K=c,∀K∈P , so we have by Lemma 11.8.4
p . c . I f d α = p . c . [ P ] f d α = K P c K α [ K ] = c K P α [ K ] = c α [ I ] p.c.∫_I fdα=p.c.∫_{[\mathbf P]} fdα=∑_{K∈\mathbf P}c_K α[K]=c∑_{K∈\mathbf P}α[K] =cα[I]

(g) The set P { J I } \mathbf P∪\{J-I\} is a partition of J J , and F F is piecewise constant on P { J I } \mathbf P∪\{J-I\} , with additional constant value c J I = 0 c_{J-I}=0 , thus
p . c . J F d α = p . c . [ P { J I } ] F d α = K P { J I } c K α [ K ] = K P c K α [ K ] + 0 α [ J I ] = K P c K α [ K ] = p . c . I f d α \begin{aligned}p.c.∫_J Fdα&=p.c.∫_{[\mathbf P∪\{J-I\}]} Fdα=∑_{K∈\mathbf P∪\{J-I\}}c_K α[K]\\&=∑_{K∈\mathbf P}c_K α[K]+0\cdot α[J-I]=∑_{K∈\mathbf P}c_K α[K]\\&=p.c.∫_I fdα\end{aligned}

( h ) We have P J = { L J : L P } \mathbf P_J=\{L∩J:L∈P\} and P K = { L K : L P } \mathbf P_K=\{L∩K:L∈P\} partitions of J J and K K , and since f f is piecewise constant with P \mathbf P , it’s easy to see f J f|_J is piecewise constant with P J \mathbf P_J on J J , f K f|_K is piecewise constant with P K \mathbf P_K on K K . As we have J I J⊆I and K I K⊆I , we define
F ( x ) = { f J ( x ) , x J 0 , x J , G ( x ) = { f K ( x ) , x K 0 , x K F(x)=\begin{cases}f|_J (x),&x∈J\\0,&x\notin J\end{cases},\quad G(x)=\begin{cases}f|_K (x),&x∈K\\0,&x\notin K\end{cases}
Since { J , K } \{J,K\} is a partition of I I , we have f = f J + f K = F + G f=f|_J+f|_K=F+G , thus use ( a ) ,( g ) we have
p . c . I f d α = p . c . I ( F + G ) d α = p . c . I F d α + p . c . I G d α = p . c . J f J d α + p . c . K f K d α \begin{aligned}p.c.∫_I fdα&=p.c.∫_I(F+G)dα=p.c.∫_I Fdα+p.c.∫_I Gdα\\&=p.c.∫_Jf|_J dα+p.c.∫_K^ f|_K dα\end{aligned}

Exercise 11.8.4

Proposition: Let I I be a bounded interval, and let f f be a function which is uniformly continuous on I I . Let α : X R α:X→\mathbf R be defined on a domain containing I I which is monotone increasing, Then f f is Riemann-Stieltjes integrable on I I with respect to α α .
Proof: From Proposition 9.9.15 we see that f f is bounded. Now we have to show that I f d α = I f d α \underline{∫}_I fdα=\overline{∫}_I fdα .
If I I is a point or the empty set then the theorem is trivial, so let us assume that I I is one of the four intervals [ a , b ] , ( a , b ) , ( a , b ] [a,b],(a,b),(a,b] , or [ a , b ) [a,b) for some real numbers a < b a<b .
Let ε > 0 ε>0 be arbitrary. By uniform continuity, there exists a δ > 0 δ>0 such that f ( x ) f ( y ) < ε |f(x)-f(y)|<ε whenever x , y I , x y < δ x,y∈I,|x-y|<δ . By the Archimedean principle, there exists an integer N > 0 N>0 such that ( b a ) / N < δ (b-a)/N<δ , Note that we can partition I I into N N intervals J 1 , , J N J_1,\dots,J_N , each of length ( b a ) / N (b-a)/N , we thus have
I f d α k = 1 N ( sup x J k f ( x ) ) α [ J k ] , I f d α k = 1 N ( inf x J k f ( x ) ) α [ J k ] \overline{\int}_I fdα≤∑_{k=1}^N\left(\sup_{x∈J_k}⁡f(x) \right)α[J_k],\quad \underline{\int}_I fdα≥∑_{k=1}^N\left(\inf_{x∈J_k}⁡f(x)\right)α[J_k]
So we have
I f d α I f d α k = 1 N ( sup x J k f ( x ) inf x J k f ( x ) ) α [ J k ] ε k = 1 N α [ J k ] = ε α [ I ] = ε ( α ( b ) α ( a ) ) \begin{aligned}\overline{∫}_I fdα-\underline{∫}_I fdα&≤∑_{k=1}^N\left(\sup_{x∈J_k}⁡f(x)-\inf_{x∈J_k}⁡f(x) \right)α[J_k] \\&≤ε∑_{k=1}^Nα[J_k] =εα[I]=ε(α(b)-α(a))\end{aligned}
This shows I f d α I f d α \underline{∫}_I fdα-\overline{∫}_I fdα cannot be positive.

Exercise 11.8.5

We let ε > 0 ε>0 be arbitrary. Then since f f is continuous, δ > 0 ∃δ>0 such that
f ( x ) f ( 0 ) < ϵ , x ( δ , δ ) |f(x)-f(0)|<ϵ,\quad ∀x∈(-δ,δ)
Since f f is continuous on [ 1 , 1 ] [-1,1] , f f is bounded, thus M > 0 ∃M>0 , s.t.
f ( x ) M , x [ 1 , 1 ] |f(x)|≤M,\quad ∀x∈[-1,1]
We let { [ 1 , δ ] , ( δ , δ ) , [ δ , 1 ] } \{[-1,-δ],(-δ,δ),[δ,1]\} be a partition of [ 1 , 1 ] [-1,1] , then the piecewise constant function
g ( x ) = { f ( 0 ) + ϵ , x ( δ , δ ) M , x [ 1 , 1 ] \ ( δ , δ ) g(x)=\begin{cases}f(0)+ϵ,&x∈(-δ,δ)\\M,&x∈[-1,1]\backslash (-δ,δ)\end{cases}
majorizes f f and we have
[ 1 , 1 ] f d sgn [ 1 , 1 ] g d sgn = M ( sgn ( δ ) sgn ( 1 ) ) + ( f ( 0 ) + ϵ ) ( sgn ( δ ) sgn ( δ ) ) + M ( sgn ( 1 ) sgn ( δ ) ) = 2 ( f ( 0 ) + ϵ ) \begin{aligned}&\quad\quad\overline{∫}_{[-1,1]} fd\text{sgn}≤∫_{[-1,1]}gd\text{sgn}\\&=M(\text{sgn}(-δ)-\text{sgn}(-1))+(f(0)+ϵ)(\text{sgn}(δ)-\text{sgn}(-δ))+M(\text{sgn}(1)-\text{sgn}(δ))\\&=2(f(0)+ϵ)\end{aligned}
Similarly, the piecewise constant function
h ( x ) = { f ( 0 ) ϵ , x ( δ , δ ) M , x [ 1 , 1 ] \ ( δ , δ ) h(x)=\begin{cases}f(0)-ϵ,&x∈(-δ,δ)\\-M,&x∈[-1,1]\backslash(-δ,δ) \end{cases}
minorizes f f and we have
[ 1 , 1 ] f d sgn [ 1 , 1 ] h d sgn = 2 ( f ( 0 ) ϵ ) \underline{∫}_{[-1,1]} fd\text{sgn}≥∫_{[-1,1]}hd\text{sgn}=2(f(0)-ϵ)
Thus we have
2 ( f ( 0 ) ϵ ) [ 1 , 1 ] f d sgn [ 1 , 1 ] f d sgn 2 ( f ( 0 ) + ϵ ) 2(f(0)-ϵ)≤\underline{∫}_{[-1,1]} fd\text{sgn}≤\overline{∫}_{[-1,1]}fd\text{sgn}≤2(f(0)+ϵ)
And the result follows.

发布了77 篇原创文章 · 获赞 14 · 访问量 2791

猜你喜欢

转载自blog.csdn.net/christangdt/article/details/103953602