$$ f_n(x) = \left\{ \begin{array}{ll} n & \mbox{if $0 \lt x \lt \frac{1}{n}$};\\ 0 & \mbox{otherwise}.\end{array} \right. \ $$

I have to show that

  1. $\lim_{n \to \infty}f_n(x)$ exists in terms of distributions

  2. $\lim_{n \to \infty}f_n^2(x)$ doesn't exist in terms of distributions.

  3. $\lim_{n \to \infty}(f_n^2(x)-n\delta)$ exists in terms of distributions

For $(1)$ $$n\int_{0}^{\frac{1}{n}}\phi(x) dx \to \phi(0) , \text{for}, n \to \infty$$

For $(2)$ $$n^2\int_{0}^{\frac{1}{n}}\phi(x) dx $$ Now there exists a $C^{\infty}$ cut off function such that $\phi(x) =1$ for $x \in (0,\frac{1}{n+1})$ and $\phi(x)=0, x \gt \frac{1}{n}$. For this particular choice of $\phi$ this integral goes to $\infty$.

For $(3)$ This will be $$\int_{-n}^{\frac{1}{n}} -(n\delta)\phi(x) dx +(n^2-n\delta)\int_{0}^{\frac{1}{n}}\phi(x) dx+\int_{\frac{1}{n}}^{n} -(n\delta)\phi(x) dx $$

I am unable to see where this would converge.

Thanks for the help!!


A quick remark on your solution of (2): You need to say that $\phi = 1$ on $[0;1]$ or something like that, as $\phi$ has to be independent of $n$. I know it's pedantic, but this is how you can loose marks on exams.

For (3): you actually want to show that the map $$T(\phi) = \lim_{n \to \infty} \int_{0}^{\frac{1}{n}} n^2f(x)dx - n \delta(\phi)$$ is a distribution. Because $\delta(\phi) = \phi(0) = \int_{0}^{\frac{1}{n}}n\phi(0)dx$, the above map is actually $T(\phi) = n^2\int_{0}^{\frac{1}{n}}\phi(x) - \phi(0) dx$ There are many ways to find this limit. I will use Taylor's theorem: $\phi(x) \phi(0) + x \phi'(0) + O(n^{-2})$ where I used that in the integral $|x| \leq \frac{1}{n}$. The remainder term is bounded by $\frac{C}{n^2}$ so when you multiply by $n^2$ and integrate from $0$ to $\frac{1}{n}$, this will converge to $0$. So, $$ T(\phi) = \lim n^2\int_{0}^{\frac{1}{n}} x \phi'(0) dx = \frac{\phi'(0)}{2}$$ Indeed, this is a distribution: it is a linear functional and it is continuous: if $\phi_k \to \phi$ in the topology of $D$ then $\sup_{\mathbb{R}} |\phi_k' - \phi'|\to 0$, i.e $T(\phi_k) \to T(\phi)$. If the last explanation looks meaningless you might use a different definition of the topology of $\mathbb{D}$: you can look at Distributions for more information.

I hope this helps.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.