Proof of normed vector space not metrizable in the weak topology?

67 Views Asked by At

I'm trying to prove that the weak topology of an infinite dimensional normed space is not metrizable. Here is what I have come up with, I'm not sure if it is correct, please correct me if it wrong! Thank you! Here, I used $Z(f)$ to denote the kernel of a linear functional $f$, and $'$ to denote the dual space.

Assume that $X$ is infinite dimensional, let $Y$ be the completion of $X$, so $Y$ is a Banach space, $\tau:X\rightarrow Y$ is a normed preserving linear map with dense image. A net $(x_n)\in X$ converges to $x\in X$ in the weak topology of $X$ iff $f(x_n)\rightarrow f(x)$ for all $f\in X'$ iff $g(\tau(x_n))\rightarrow g(\tau(x))$ for all $g\in X'$ iff $\tau(x_n)\rightarrow\tau(x)$ in the weak topology of $Y$. Therefore, the weak topology of $X$ agrees with the subsapce topology of the range of $\tau$ as a subspace of the weak topology of $Y$. Hence if we can show the weak topology of $Y$ is not metrizable, then the weak topology of $X$ is also not metrizable.

Therefore, we can assume that $X$ is an infinite dimensional Banach space, so its dimension must be uncountable. If $X$ is metrizable under the weak topology, then $X$ is first countable at $0$ under the weak topology. This implies that we can find a neighborhood basis of the weak topology of $X$ at $0$, that consists of a sequence of open sets $(U_n)_1^\infty$ each one is of the form $\bigcap_{i=1}^j f_i^{-1}(B(0,\epsilon))$, where $f_i\in X'$ is nonzero and $\epsilon>0$, $j\in\mathbb{Z}^+$. Let $\mathcal{F}$ be the collection of all such $f_i$'s for for all $U_n$'s. $\mathcal{F}$ is a countable set. Since $X$ is of uncountable dimension, it is but a linear algebraic fact that $\bigcap_{f\in\mathcal{F}}Z(f)$ is nonzero. Hence, we can find a nonzero $v\in\bigcap_{f\in\mathcal{F}}Z(f)$. Using the Hahn-Banach extension theorem, we can find some $f\in X'$ such that $f(v)=1$. Then $v\in U_n$ for each $n\in\mathbb{Z}^+$ but $v\notin f^{-1}(B(0,1/2))$. However, $f^{-1}(B(0,1/2))$ is open in the weak topology of $X$, so $U_n\subset f^{-1}(B(0,1/2))$ for some $n\in\mathbb{Z}^+$. Contradiction!