Can someone help me with this question?
Let $f \in X = L^2(0, \infty)$, and define \begin{equation} (Tf)(x) = \frac{1}{x} \int_0^x f(t) dt \ . \end{equation} Show that $T$ is Bounded, and is NOT Compact.
Thanks in advance.
Edit: For boundedness, I tried: \begin{eqnarray} \| f \| \leq 1 \Rightarrow \| Tf \|^2 &=& \int_0^{\infty} \lvert (Tf)(x) \rvert ^2 \ dx \\ &=& \int_0^{\infty} \lvert \frac{1}{x} \int_0^x f(t) \ dt \rvert ^2 \ dx \\ &\leq& \int_0^{\infty} \frac{1}{x^2} \int_0^x \lvert f(t) \rvert ^2 \ dt \ dx \\ &\leq& \int_0^{\infty} \frac{1}{x^2} \| f \| ^2 \ dx \\ &=& \| f \| ^2 \int_0^{\infty} \frac{1}{x^2} \ dx \\ \end{eqnarray} But the problem is that $\int_0^{\infty} \frac{1}{x^2} \ dx$ is NOT finite.
To prove that $T$ is not compact, I think we should give a sequence of functions, $\{ f_n \}_{n=1}^{\infty}$, such that for every $n$, we have $\| f_n \| \leq 1$ and $\{ Tf_n \}$ has no Cauchy subsequence. But I can't find such sequence of functions.
I thought I would just offer an alternate solution using Minkowski's integral inequality applied to:
$$ \| Tf \|_2 = \left\{ \int_0^{\infty} \left( \int_0^x x^{-1} f(t)\,dt \right)^{2}\,dx \right\}^{1/2} $$
But first we do a variable substitution $t \rightarrow xt$ so the inner integral is integrating over a fixed space:
$$ \| Tf \|_2 = \left\{ \int_0^{\infty} \left( \int_0^1 f(xt)\,dt \right)^{2}\,dx \right\}^{1/2} $$
Minkowski's inequality then yields:
$$ \| Tf \|_2 \leq \int_0^1 \left \{ \int_0^{\infty} f^2(xt)\, dx \right\}^{1/2}\, dt $$
Now undo our variable substitution, this time sending $x \rightarrow x/t$, and get:
$$ \| Tf \|_2 \leq \int_0^1 \left \{ \int_0^{\infty} t^{-1} f^2(x)\, dx \right\}^{1/2}\, dt = \left( \int_0^1 t^{-1/2}\,dt \right)\left ( \int_0^{\infty} f^2(x)\,dx\right )^{1/2}$$
And, of course, $\int_0^1 t^{-1/2}\,dt$ is finite.
To answer your second question. Let
$$ f(x) = \begin{cases} & 0 & x \leq 0 \\ & 1 & 0 < x < 1/2 \\ & -1 & 1/2 \leq x \leq 1 \\ & 0 & 1 \leq x \end{cases} $$
So $\|f\|_2 = 1$. Now let $f_n(x) = n^{-1/2}f\left(\frac{x - n}{n}\right)$. So,
$$\|f_n\|_2 = n^{-1/2} \left( \int_n^{2n} dt \right)^{1/2} = 1.$$
The important property here is that $\int_0^1 f(t) dt = 0$. This is going to ensure that $Tf_n$ have disjoint support. Why? Recall:
$$ Tf_n(x) = \frac{1}{x}\int_0^x n^{-1/2} f\left(\frac{t-n}{n}\right)\,dt $$
Since $f(x)$ is $0$ for $x < 0$ we have that $Tf_n$ is 0 for $x < n$. And since $\int_0^1 f(t)\,dt =0 $, we have:
$$ Tf_n(2n) = \frac{1}{2n} \int_n^{2n} f\left(\frac{t - n}{n}\right)\,dt = \frac{1}{2n} \int_0^n f\left(\frac{t}{n}\right)\,dt = 0$$
And you can see that $Tf_n(x) = 0$ for $x > 2n$. So if we take $f_{2^k}$ as $k = 1, 2, 3, \dots$ we can see that $Tf_{2^k}$ have disjoint supports. So to prove that these have no convergent subsequence you just need to show that $\|Tf_k\|_2$ is bounded below for any integer $k$. We will estimate $\| Tf_k \|_2^2$ but only focus on those x where $f_k$ is positive.
$$\| Tf_k \|_2^2 = \int_k^{2k} \frac{1}{x^2} \left\{ \int_0^x f_k(t)\,dt \right\}^2\,dt \geq \int_k^{3k/2} \frac{1}{4k^2} \left\{\int_k^x f_k(t)\,dt \right\}^2\,dx $$
We have:
$$ \int_0^x f_k(t)\,dt = k^{-1/2}(x - k) $$
for $k \leq x \leq 3k/2$. So we get:
$$\| Tf_k \|_2^2 \geq \frac{1}{4k^2} \frac{1}{k} \int_k^{3k/2} (x - k)^2\, dx$$
And this can be shown to be bounded below be something like $1/100$. Sorry, but nothing else comes to mind for proving this.