It is well known that, at least as sets, $$\mathcal{S}'(\mathbb{R}^n)=\bigcup_{m\in\mathbb{N}}(1+|x|^2)^mH^{-m}(\mathbb{R}^n)$$ where $\mathcal{S}'(\mathbb{R}^n)$ is the space of tempered distribution (i.e. the dual of Schwartz space), $H^{-m}(\mathbb{R}^n)$ denotes the $L^2$-Sobolev space of order $-m$ with norm $\Vert u\Vert^2=\int_{\mathbb{R}^n}(1+|\xi|^2)^{-m}|\hat{u}(\xi)|^2\,d\xi$ and $(1+|x|^2)^mH^{-m}(\mathbb{R}^n)$ is the weighted Sobolev space $$(1+|x|^2)^mH^{-m}(\mathbb{R}^n)=\{u\in\mathcal{S}'(\mathbb{R}^n): (1+|x|^2)^{-m}u\in H^{-m}(\mathbb{R}^n)\}$$
I also read somewhere that the strong topology on $\mathcal{S}'$ is the strict inductive limit topology induced by the RHS, however, I have not been able to find a proof of this fact.
I would be content with a weaker sequential statement: I would like to prove that $\{u_n\}\subset\mathcal{S}'(\mathbb{R}^n)$ converges weakly to $u$ in $\mathcal{S}'(\mathbb{R}^n)$ (i.e. $u_n(\phi)\rightarrow u(\phi)$ in $\mathbb{C}$ as $n\rightarrow\infty$ for all $\phi\in\mathcal{S}(\mathbb{R}^n)$) if and only if there is some $m\in\mathbb{N}$ for which $u_m\rightarrow u$ in the topology of $(1+|x|^2)^mH^{-m}(\mathbb{R}^n)$.
Does anyone know a proof of this fact?
Taking the real part of everything we can restrict to real-valued Schwartz functions and real valued distributions.
Let the order of a tempered distribution be the least positive integer $m$ such that it extends to a continuous functional on the Hilbert space $B_m=(1+|x|)^{-m} H^m$
Let $c_k = \sup_{m\le k}\|f_m\|_{\textstyle (B_m)'}$
Construct a sequence $\phi_k\in S$ such that $$\|\phi_k\|_{B_{k-1}} \le \frac1{c_k} 2^{-k}\ \ and \ \ \langle f_k,\phi_k\rangle > e^k+\sum_{m<k} |\langle f_k,\phi_m\rangle|$$ (by definition of the order this is always possible)
$$\Phi= \sum_k \phi_k \ converges \ in \ all \ B_m \ thus \ in\ S$$
Also for $m>k$, $$|\langle f_k,\phi_m \rangle|\le \|f_k\|_{(B_k)'}\|\phi_k\|_{B_k}\le c_m\|\phi_m\|_{B_{m-1}}\le 2^{-m}$$
Thus
$$\langle f_k,\Phi \rangle = \sum_{m<k} \langle f_k,\phi_m\rangle+\langle f_k,\phi_k\rangle+ \sum_{m>k} \langle f_k,\phi_m\rangle$$ $$ > e^k-\sum_{m>k} 2^{-m}$$ And hence $$\sum_k \langle f_k,\Phi\rangle = \infty$$ which is a contradiction.