Is the composition of an $L^2$-function with a smooth one again an $L^2$-function?

182 Views Asked by At

There's that claim in the book I'm reading that I don't manage to prove.

We consider the function

$$ b(x)= \int_0^1 H(x(t))dt.$$

Where $x \in C^{\infty}(S^1,\mathbb{R}^n)$ and $H : \mathbb{R}^n \rightarrow \mathbb{R}$ is smooth.

We know that the function $H$ is such that

(1) $H(x) \equiv 0$ near $x=0$ and

(2) $H(x)= (\pi + \epsilon)q(x)$, for large values of $\mid x \mid$. Here $q(x)$ is a quadratic form.

Furthermore the following estimates hold for all $x \in \mathbb{R}$:

(3) $\mid H(x) \mid \leq M \mid x \mid^2$,

(4) $\mid d^2H(x) \mid \leq M.$

For a positive constant $M$.

The claim is that $b$ is defined for all $x \in L^2(S^1,\mathbb{R}^n)$. I am not sure what I should show here. This is what I've done

Let $x \in L^2 (S^1,\mathbb{R}^n)$, then by definition we have

$$ \int_0^1 \mid x(t) \mid^2 dt < \infty. $$

Now compute the value of $b(x)$ and use (3):

$$b(x)=\int_0^1 H(x(t)) dt \leq \int_0^1 M \mid x(t) \mid^2 dt = M \int_0^1 \mid x(t) \mid^2 dt < \infty.$$

I would argue that $b(x)$ cannot diverge to $- \infty$ because of (1) and (2).

Please let me know if I really missed something. Thank you very much