Contradiction regarding the continuity of a distribution

48 Views Asked by At

Consider $ T \in\mathcal{D}'(\mathbb{R})$ defined as follows, $$\phi \to T(\phi) = \sum_{n=-\infty}^{+\infty} \phi(n)$$

Next, let $\psi \in \mathcal{D}(\mathbb{R})$ be a function with support in $[-1,1]$ such that $0 \le \psi(x) \le 1$ for all $x \in \mathbb{R}$ and such that $\psi(x) = 1$ for all $x \in [-0.5,0.5]$. Next, consider the sequence $$\phi_n(x) = \frac{1}{\sqrt{1+n}}\psi(\frac{x}{n})$$ Note that $\phi_n$ converges uniformly to zero as $n \to \infty$, however, $T(\phi_n) \to \infty$. Does this contradict the continuity of $T$?

My attempt:

Since the question states $T$ is a distribution, it must be continuous and so I'm led to believe that I must show why this isn't a contradiction.

My thoughts are that $\phi_n \notin \mathcal{D}(\mathbb{R})$, and so $T(\phi_n)$ not converging to zero isn't a contradiction. I believe that for each $n$, $\phi_n(x)$ is in $C^\infty$ and smooth, as $\phi^{(k)}_n(t) = \frac{1}{n^k}\frac{1}{\sqrt{1+n}}\psi^{(k)}(\frac{x}{n})$ and $\psi$ is in $\mathcal{D}(\mathbb{R})$ and the terms in front of it will simply be a constant.

However, we can also note that the support of $\phi_n$ is given by $[-n,n]$. Thus, as $n \to \infty$, the support of $\phi_n$ also goes to infinity and hence, in the limit $\phi_n$ does not have compact support and so is not in $\mathcal{D}(\mathbb{R})$.

I'm not too confident in my answer, but I've been reviewing this question for awhile and can't think of anything else. Does anyone have any ideas?

1

There are 1 best solutions below

4
On BEST ANSWER

Uniform pointwise convergence is certainly necessary for convergence of $\varphi_n$ in test functions, but it is not sufficient. One further requirement of a Cauchy sequence (and also Cauchy net!) is that the supports be (uniformly) bounded. A sequence such as you constructed definitely does not have this property. So your $\varphi_n$ is not a Cauchy sequence in the space of test functions, so there's no paradox here.