User Tools

Site Tools


math105-s22:notes:lecture_12

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Next revision
Previous revision
math105-s22:notes:lecture_12 [2022/02/23 23:02]
pzhou created
math105-s22:notes:lecture_12 [2022/02/25 00:35] (current)
pzhou
Line 4: Line 4:
  
   * Redo Fubini's Theorem (Tao 8.5.1)   * Redo Fubini's Theorem (Tao 8.5.1)
 +  * Vitali Covering Lemma. 
  
 ===== Upper and Lower Lebesgue integral ===== ===== Upper and Lower Lebesgue integral =====
Line 31: Line 32:
 F+(x)dx=F(x)dx \int F_+(x) dx = \int F_- (x) dx  F+(x)dx=F(x)dx \int F_+(x) dx = \int F_- (x) dx
 hence F+(x)=F(x)F_+(x) = F_-(x) for almost all xx. Thus, for a.e. xx, we have $\uint f(x,y) dy = \lint f(x,y) dy$, thus f(x,y)dy\int f(x,y) dy exists for a.e. x.  hence F+(x)=F(x)F_+(x) = F_-(x) for almost all xx. Thus, for a.e. xx, we have $\uint f(x,y) dy = \lint f(x,y) dy$, thus f(x,y)dy\int f(x,y) dy exists for a.e. x. 
 +
 +==== A Lemma ====
 +Suppose AA is measurable, and BAB \In A any subset, with Bc=A\BB^c = A \RM B. Then
 +m(A)=m(B)+m(Bc) m(A) = m^*(B) + m_*(B^c)
 +Proof: 
 +m(B)=inf{m(C)ACB,Cmeasurable}=inf{m(A)m(Cc)ACB,Cmeasurable} m^*(B) = \inf \{ m(C) \mid A \supset C \supset B, C \text{measurable} \} = \inf \{ m(A) - m(C^c) \mid A \supset C \supset B, C \text{measurable} \}
 +=m(A)sup{m(Cc)ACB,Cmeasurable}=m(A)sup{m(Cc)CcBc,Ccmeasurable}=m(A)m(Bc) = m(A) - \sup \{ m(C^c) \mid A \supset C \supset B, C \text{measurable} \} = m(A) - \sup \{ m(C^c) \mid C^c \subset B^c, C^c \text{measurable} \} = m(A) - m_*(B^c)
 +
 +===== Pugh 6.8: Vitali Covering =====
 +\gdef\vcal{\mathcal{V}}
 +A Vitali covering V\vcal of a set ARnA \In \R^n is such that, for any pA,r>0p \in A, r > 0, there is a covering set VVV \in \vcal, such that {p}VBr(p)\{p\} \subsetneq V \In B_r(p), where Br(p)B_r(p) is the open ball of radius rr around pp.
 +
 +**Vitali Covering Lemma:** Let V\vcal be a Vitali covering of a measurable bounded subset AA by closed balls, then for any $\epsilon>0$, there is a countable disjoint subcollection $\vcal' = \{V_1, V_2, \cdots \},suchthat, such that A \RM \cup_k V_kisanullset,and is a null set, and \sum_k m(V_k) \leq m(A) + \epsilon$. 
 +
 +Proof: The construction is easy, like a 'greedy algorithm'. First, using the given ϵ\epsilon, we find an open subset WAW \supset A, with m(W)m(A)+ϵm(W) \leq m(A) + \epsilon. Let V1={VV:VW}\vcal_1 = \{V \in \vcal: V \In W\}, and d1=sup{diamV:VV1}d_1 = \sup \{diam V: V \in \vcal_1\}. We pick V1V1V_1 \in \vcal_1 where the diameter is sufficiently large, say diamV1>d1/2diam V_1 > d_1 /2. Then, we delete V1V_1 from WW, let W2=W\V1W_2 = W \RM V_1, and consider V2={VV1,VW2}\vcal_2 = \{ V \in \vcal_1, V \In W_2\}, and define d2=sup{diamV:VV2}d_2 = \sup \{diam V: V \in \vcal_2\}, and pick V2V_2 among V2\vcal_2 so that diamV2>d2/2diam V_2 > d_2 /2. Repeat this process, we get a collection of disjoint closed balls {Vi}\{V_i\}. Suffice to show that A\ViA \RM \cup V_i is a null set. 
 +
 +The crucial claim is the following, for any positive integer NN, we have
 +k=N5VkA\(i=1N1Vi) \cup_{k=N}^\infty 5 V_k \supset A \RM (\cup_{i=1}^{N-1} V_i)
 +Suppose not, and there is a point aA\(i=1N1Vi)a \in A \RM (\cup_{i=1}^{N-1} V_i), but not in k=N5Vk\cup_{k=N}^\infty 5 V_k, then we can find a closed ball BVNB \in \vcal_N, such that aBa \in B. Since a5VNa \notin 5V_N, we have B⊄5VNB \not\subset 5V_N. This implies BVN=B \cap V_N = \emptyset. Draw a picture. This implies BVN+1B \in \vcal_{N+1}. Then, repeat the above story NN replaced by N+1N+1 and same a,Ba,B, we can keep going and show that BVkB \in \vcal_k for all kNk \geq N. That cannot be true, since dk0d_k \to 0, but BB has fixed radius. 
 +
 +
 +
math105-s22/notes/lecture_12.1645686128.txt.gz · Last modified: 2022/02/23 23:02 by pzhou