Want to prove that limit of integrals diverges

  • Context: Graduate 
  • Thread starter Thread starter jostpuur
  • Start date Start date
  • Tags Tags
    Integrals Limit
Click For Summary

Discussion Overview

The discussion revolves around the conditions under which the limit of integrals of a sequence of functions diverges, particularly in the context of pointwise convergence to a non-integrable function. Participants explore various scenarios involving measure spaces and the implications of different types of convergence.

Discussion Character

  • Exploratory
  • Technical explanation
  • Debate/contested
  • Mathematical reasoning

Main Points Raised

  • One participant proposes that if a sequence of functions \( f_n \) converges to a function \( f \) and the integral of \( f \) diverges, then it may be deduced that the limit of the integrals of \( f_n \) also diverges.
  • Another participant argues that uniform convergence on the interval of integration would guarantee this result.
  • A different viewpoint references Fatou's lemma, suggesting that for complex functions, if the integral diverges, at least one part of the function must diverge as well, leading to a potential conclusion about the limits of the integrals of the real parts of \( f_n \).
  • One participant raises a more complex scenario involving measure spaces, questioning whether the integrals of \( f_n \) could converge due to cancellation effects, despite the divergence of the integral of \( f \).
  • A later reply introduces a counter-example involving the function \( f(x) = \sin(x)/x \) and a sequence of functions \( f_n(x) \) that converge pointwise to \( f \) but for which the limit of the integrals converges to a finite value, challenging the initial assumptions.

Areas of Agreement / Disagreement

Participants express differing views on the conditions necessary for the divergence of the limit of integrals, with some supporting the idea that uniform convergence is sufficient, while others present counter-examples that suggest the situation is more nuanced and unresolved.

Contextual Notes

The discussion highlights limitations regarding the assumptions about convergence types and the specific properties of the functions involved, as well as the implications of measure theory in the context of integrals.

jostpuur
Messages
2,112
Reaction score
19
Suppose we know that [itex]f_n\to f[/itex] when [itex]n\to\infty[/itex] (in some sense), and that

[tex] \int f(x)dx[/tex]

diverges. Can it be deduced that

[tex] \lim_{n\to\infty} \int f_n(x)dx[/tex]

diverges too?
 
Physics news on Phys.org
If the convergence of the sequence is uniform on the interval of integration then yes
 
I see now that this seems to follow from the Fatou's lemma

I thought first it wouldn't be relevant, because I wanted to consider complex functions, but if [itex]f[/itex] is complex, and the integral diverges, then it means that at least one positive or negative part, of real or imaginary part, diverges.

Then, for example if the [itex](\textrm{Re}(f))^+[/itex] diverges, then

[tex] \lim_{n\to\infty} \int \textrm{Re}(f_n(x))^+ dx = \underset{n\to\infty}{\textrm{lim inf}} \int \textrm{Re}(f_n(x))^+ dx \geq \int \textrm{Re}(f(x))^+ dx = \infty[/tex]

I guess that's it.
 
But I don't think that my original problem would be solved yet, because my original problem looks more like this:

Suppose [itex]X[/itex] is some measure space such that [itex]\mu(X)=\infty[/itex] and [itex]X_1\subset X_2\subset\cdots[/itex] is some subset sequence such that [itex]\mu(X_k)<\infty[/itex] for all k and such that

[tex] \bigcup_{k=1}^{\infty} X_k = X[/tex]

No I know that point wisely [itex]f_n(x)\to f(x)[/itex], and that

[tex] \int\limits_X |f(x)| d\mu(x) =\infty[/tex]

and I want to prove that

[tex] \lim_{n\to\infty} \lim_{k\to\infty} \int\limits_{X_k} f_n(x) d\mu(x)[/tex]

diverges.

So is it certain, that the integrals of [itex]f_n[/itex] could not start converging for cancellation reasons?
 
I've been "writing while thinking" so it could be that my posts are little confused.
 
First, since for each n, the map

[tex]\nu(A)=\int_Af_nd\mu[/tex]

is a measure, we have by continuity, that

[tex] \lim_{n\to\infty} \lim_{k\to\infty} \int\limits_{X_k} f_n(x) d\mu(x)=\lim_{n\to\infty}\int_Xf_nd\mu[/tex]

So your question amounts to: is there a sequence of integrable functions f_n that converge pointwise to a non integrable function f but such that [itex]\lim_n\int_Xf_n[/itex] converges.

The answer is provided by the old example of X=R, f(x)=sin(x)/x. It is fairly easy to show that [itex]\int_{\mathbb{R}}|f|=+\infty[/itex] by finding an appropriate sequence of step function bounded above by |f| and whose area makes up a diverging series.

But it is known that the sequence of (integrable) functions

[tex]f_n(x):=\chi_{[-n,n]}\frac{\sin(x)}{x}[/tex]

(which converge pointwise to f) are such that

[tex]\lim_{n\to\infty}\int_{\mathbb{R}}f_n=\pi[/tex]

(more difficult).

So, here is your counter-example.
 

Similar threads

  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 17 ·
Replies
17
Views
6K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 11 ·
Replies
11
Views
3K
  • · Replies 11 ·
Replies
11
Views
3K
  • · Replies 5 ·
Replies
5
Views
3K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 1 ·
Replies
1
Views
1K