1
$\begingroup$

I call $V_{x_0,\phi,\epsilon}:=\{x:\ |\phi(x-x_0)|<\epsilon\}$, for $x_0\in X,\ \phi\in X^*,\ \epsilon>0$. Here $X^*$ denotes the space of all bounded linear functionals on $X$.

We donte $\tau_{\lVert\cdot\rVert}=$ norm topology and $\tau=$ topology generated by each $V_{x_0,\phi,\epsilon}$

Then $\displaystyle{V_{x_0,\phi,\epsilon}=\phi^{-1}(\phi(x_0)-\epsilon,\phi(x_0)+\epsilon)\in \tau_{\lVert\cdot\rVert}}$

Hence, $\tau\subseteq \tau_{\lVert\cdot\rVert}$

But I cannot prove the other way round i.e.$\tau_{\lVert\cdot\rVert}\subseteq \tau$ i.e. any open ball $B(x_0,\epsilon)$ in $\tau_{\lVert\cdot\rVert}$ is in $\tau$.

Can anyone help me to finish the part? Thanks for help in advance.

$\endgroup$

2 Answers 2

2
$\begingroup$

If $V$ is infinite-dimensional, this does not generate the norm topology. It generates the so-called weak topology on $V$.

Indeed, assume it does generate the norm topology, and consider the unit ball $B(0,1)$ which is norm-open. Then by our assumption, there exist $x_1, \ldots, x_n \in V$, $\phi_1, \ldots, \phi_n \in V^*$ and $\varepsilon_1, \ldots, \varepsilon_n > 0$ such that $$0 \in \bigcap_{k=1}^n\{x \in V : |\phi_k(x-x_k)|<\varepsilon_k\} \subseteq B(0,1).$$

Since $0$ is inside, in particular we have $|\phi_k(x_k)| < \varepsilon_k$ for all $1 \le k \le n$.

Since $V$ is infinite-dimensional, it cannot be $\bigcap_{k=1}^n \ker\phi_k = \{0\}$ (in fact, this intersection must be infinite-dimensional as well) so we can pick $y \ne 0$ such that $\phi_k(y)=0$ for all $1 \le k \le n$. Then for every $\lambda \in \Bbb{R}$ we have $$|\phi_k(\lambda y-x_k)| = |\phi_k(x_k)| <\varepsilon_k, \quad \text{ for all } 1\le k \le n$$ so $$\Bbb{R}y \subseteq \bigcap_{k=1}^n\{x \in V : |\phi_k(x-x_k)|<\varepsilon_k\} \subseteq B(0,1).$$ This is clearly a contradiction.

$\endgroup$
3
  • $\begingroup$ Okay I have misunderstood the question. Can I show that the vector addition and scalar multiplication is continuous in this weak topology? $\endgroup$ Commented Jun 25, 2021 at 18:28
  • $\begingroup$ @DeltaEpsilon Yes, you can. For addition try to show that $$+\left(V_{x_0,\phi,\frac\varepsilon2} \times V_{y_0,\phi,\frac\varepsilon2}\right) \subseteq V_{x_0+y_0, \phi,\varepsilon}$$ and then think about what if $x_0+y_0 \subseteq V_{z_0,\phi,\varepsilon}$. Similar for scalar multiplication. $\endgroup$ Commented Jun 25, 2021 at 18:52
  • $\begingroup$ Okay I understand. Thanks. Basically, I'm absolutely new to this topic, so maybe I'm asking some basic questions. $\endgroup$ Commented Jun 25, 2021 at 19:04
0
$\begingroup$

If $\mathcal{B}$ is a base of nieighbourhoods of zero for a topology $\tau$ then every set $A\in \mathcal{B}$ contains a linear subspace of finite codimension an hence cannot be a norm bounded unless the space $X$ is finite dimensional. Therefore the two topologies are not equivalent if $X$ is not finite dimensional.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .