作者ppia (papayaPaul)
看板Math
標題[分析] Limits of Measureable Functions
時間Sun Jun 27 18:00:30 2010
在 Daniell integral (
http://en.wikipedia.org/wiki/Daniell_integral ) 中,
可測函數定義有以下兩種定義﹔
(a) f is measureable iff for any positive integrable g
+ -
min(f ,g) - min(f ,g) = mid(g,f,-g) is integrable.
(b) f is measureable iff f is a limit a.e.
of a sequence of elementary functions.
一般來說(b)=>(a)但反之不亦然, 在下列條件底下, (a)=>(b)
(*) There exists an increasing sequence of elementary functions φ_n with
lim φ_n > 0 a.e..
n→∞
我要問的是:
"(a.e.) limits of sequences of measureable functions is also measureable."
這件事對(a)成立,那麼一般來說, 在(*)不成立的狀況下, 上述是否也對(b)成立?
--
※ 發信站: 批踢踢實業坊(ptt.cc)
◆ From: 125.231.229.38
自問自答... 答案是肯定的
Let s_n be a sequence of measureable(b) functions tending a.e. to a
function f. Then f is a measureable function(a). Besides, since each s_n is
a.e. the limit of a sequence of elementary functions {σ_mn︱m=1,2,...},
if we let σ_n to be an enumeration of the double sequence {|σ_mn|} and let
φ_n = max(σ_1, σ_2, ...,σ_n),
lim φ_n > 0 a.e. on E:= supp(f).
n→∞
Put f_n = mid( nφ_n, f, -nφ_n ) → f a.e. and g_n = f_[n+1] - f_n. Then
each g_n is integrable with Σg_n = f a.e.. Since each g_n is integrable, there
corresponds a sequence of elementary functions {ψ_kn︱k=1,2,...} that tends to
g_n a.e. with
n+k
∫|ψ_kn-g_n| dμ < 1 / 2 .
Put ψ_n = ψ_n1 + ψ_n2 + ... +ψ_nn. It remains to show that ψ_n→f a.e..
Let E be a measure zero set off which
∞
Σ g = f and lim ψ = g hold.
n=1 n k→∞ kn n
Put E_n = {x︱|f(x)| < nφ_n} - E. Fix some N, for each x in E_N, g_n(x) = 0
for all n > N. Hence
ψ (x) + ψ (x) + ... + ψ (x)
n1 n2 nN
—→ g(x) + g(x) + ... + g(x) = f(x) as n →∞.
1 2 N
Put ζ_n = ψ_n - (ψ_n1 + ψ_n2 + ... +ψ_nN).
│ζ_n│ = │ψ_n N+1 + ... +ψ_nn│ ≦ │ψ_n N+1│ + ... +│ψ_nn│
Since g_n = 0 on E_N for n>N,
∫ │ζ_n│dμ ≦ ∫ │ψ - g │+ ... + │ψ - g│dμ
E_N E_N n N+1 N+1 nn n
n
≦ ∫│ψ - g │+ ... + │ψ - g│dμ ≦ 1/2
n N+1 N+1 nn n
By Beppo Levi Lemma, Σζ_n converges absolutely a.e. on E_N, which implies
ζ_n→0 a.e. on E_N. Hence ψ_n→f a.e. on each E_N.
※ 編輯: ppia 來自: 125.231.229.38 (06/27 22:22)
推 Vulpix :為什麼g_n可積啊? 06/28 00:09
→ ppia :因為f_n可積 06/28 00:10
推 Vulpix :推這篇,Daniel int.也是一個切入實分析的好方向 06/28 00:19
→ ppia :很少實分析課本用non-measure approach來介紹積分 06/28 00:23
→ ppia :Daniell integral放在泛函理論裡面會顯得比較自然 06/28 00:24
推 Vulpix :當然一開始就從可積分切入不太自然,可是我覺得 06/28 00:26
推 Vulpix :能用其他角度看實分析很重要,說明不是只有從measure 06/28 00:29
→ Vulpix :開始這唯一一條路 06/28 00:29
推 XCHCH2 :那這跟一般測度論裡面的可測函數定義等價嗎? 06/28 00:34
推 XCHCH2 :f是可測函數iff f^-(a,∞)是可測集,for all a 06/28 00:37
推 Vulpix :是等價的,不過看起來一開始選的基本函數積分不是 06/28 00:39
→ Vulpix :習慣的積分的話,也可以構造出其他測度耶 06/28 00:40
→ ppia :其實滷乙烯大大的問題比較不那麼直接 06/28 00:43
→ ppia :這個問題的癥結在於 兩種理論給的可測集是否一樣 06/28 00:43
→ ppia :如果是以(a)當作可測函數的定義的話 那不必然 06/28 00:44
→ ppia :如果是(b)的話 那我不是很肯定 06/28 00:44
→ ppia :給定elementary sets 我們可以定義elementary fx's為 06/28 00:46
→ ppia :charateristic fx of elementary set的線性組合 06/28 00:47
→ ppia :如此構造的理論'可測'等價 至少是用(b)作為定義時 06/28 00:49
→ ppia :然而若先給定elementary fx 若elementary fx非 06/28 00:50
→ ppia :simple fx 要如何定義elementary set 這我就沒想過了 06/28 00:50
推 Vulpix :我是看到原po給的wiki連結才這樣說的,裡面提到說 06/28 00:52
→ ppia :簡單的說 若ele. fx 非 simple fx. 這個"等價"到底 06/28 00:52
→ Vulpix :measure in Daniel sense <=> usual measure 06/28 00:53
→ ppia :是什麼意思 並不是很清楚 06/28 00:53
→ ppia :那應該是在比較特殊狀況 比如說Lebesgue積分時 06/28 00:56
推 XCHCH2 :thanks 還第一次有人叫我滷乙烯XD 06/28 01:44