1. Limited time only! Sign up for a free 30min personal tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Homework Help: Real analysis: Problem similar to uniformly integrable

  1. Nov 23, 2012 #1
    1. The problem statement, all variables and given/known data

    Assume [itex]\mu[/itex](X) >0 and that f is a measurable function that maps X into ℝ and satisfies f(x) >0 for all x[itex]\in[/itex]X.

    Let [itex]\alpha[/itex] be any fixed real number satisfying 0<[itex]\alpha[/itex]<[itex]\mu[/itex](X) <infinity Prove that

    inf { [itex]\int[/itex][itex]_{E}[/itex]f d[itex]\mu[/itex] : E[itex]\in[/itex]M, [itex]\mu[/itex](E) ≥[itex]\alpha[/itex]} >0.

    (Hint. First prove for any [itex]\delta[/itex] satisfying 0<[itex]\delta[/itex]<[itex]\alpha[/itex], prov there exists n [itex]\in[/itex] N such that B[itex]_{n}[/itex] = {x:f(x)[itex]\geq[/itex]1/n} satisfies [itex]\mu[/itex](B[itex]_{n}[/itex]) [itex]\geq[/itex]
    [itex]\mu[/itex](X) - [itex]\delta[/itex]. Then prove that if [itex]\mu[/itex](E) ≥[itex]\alpha[/itex] then [itex]\mu[/itex](E[itex]\cap[/itex]B[itex]_{n}[/itex]) ≥[itex]\alpha[/itex]-[itex]\delta[/itex]

    2. Relevant equations

    Not sure what is relevant.

    3. The attempt at a solution

    So, the hint suggests the first thing I do is show that given 0<[itex]\delta[/itex]<[itex]\alpha[/itex] i can find an n so that [itex]\mu[/itex](B[itex]_{n}[/itex]) [itex]\geq[/itex] [itex]\mu[/itex](X) - [itex]\delta[/itex]. So since f(x) >0 then [itex]\mu[/itex](X) = [itex]\mu[/itex]({x:f(x)>0}) = [itex]\mu[/itex]({x:f(x)≥1/n} [itex]\cup[/itex]{x:f(x)<1/n}). These are disjoint so we have [itex]\mu[/itex]({x:f(x)≥1/n}) + [itex]\mu[/itex]({x:f(x)<1/n}). If i show the delta relation with [itex]\mu[/itex]({x:f(x)<1/n}) then i get [itex]\mu[/itex](B[itex]_{n}[/itex]) [itex]\geq[/itex] [itex]\mu[/itex](X) - [itex]\delta[/itex]. It seems really trivial though and im not sure what guarantees this.

    The next part asks you to show that [itex]\mu[/itex](E[itex]\cap[/itex]B[itex]_{n}[/itex]) ≥[itex]\alpha[/itex]-[itex]\delta[/itex]. Ok so [itex]\mu[/itex](E[itex]\cap[/itex]B[itex]_{n}[/itex]) = [itex]\mu[/itex](E) +[itex]\mu[/itex](B[itex]_{n}[/itex]) - [itex]\mu[/itex](E[itex]\cup[/itex]B[itex]_{n}[/itex])

    The following hold: [itex]\mu[/itex](E[itex]\cup[/itex]B[itex]_{n}[/itex]) ≤[itex]\mu[/itex](X) (both are subsets)
    [itex]\mu[/itex](B[itex]_{n}[/itex]) ≥ [itex]\mu[/itex](X) - [itex]\delta[/itex] (previous part)
    [itex]\mu[/itex](E)≥[itex]\alpha[/itex] (hypothesis)

    So [itex]\mu[/itex](E[itex]\cap[/itex]B[itex]_{n}[/itex]) = [itex]\mu[/itex](E) +[itex]\mu[/itex](B[itex]_{n}[/itex]) - [itex]\mu[/itex](E[itex]\cup[/itex]B[itex]_{n}[/itex]) ≥ [itex]\mu[/itex](X) - [itex]\delta[/itex] + [itex]\mu[/itex](E) - [itex]\mu[/itex](X) =[itex]\mu[/itex](E)- [itex]\delta[/itex]≥[itex]\alpha[/itex]-[itex]\delta[/itex]

    So I've done most of what the hint wants... the problem is I don't know how this helps me.
  2. jcsd
  3. Nov 23, 2012 #2
    The title is misleading, I thought it related to uniformly intagrable but I don't think it does.
  4. Nov 24, 2012 #3
    I figured out why there is an n for every δ. So I figured out both parts of the hint, but I still have no clue how to make these hints help me here. Do I have to show that

    inf {[itex]\int[/itex][itex]_{B_{n}\cap E}[/itex]f d[itex]\mu[/itex], E[itex]\in[/itex]M, [itex]\mu (B_{n}\cap E)[/itex]≥α-δ} > 0 for each n? And then as n-> infinity we have the desired result? I'm still not sure how to prove this case though...
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook