We will say that ${T_i}\subset B(X,Y^*)$ converges to $T$ in W*-operator topology if $T_i(x)\rightarrow T(x)$ in W*-topology of $Y^*$( $\forall y\in Y \langle T_i(x),y\rangle \rightarrow \langle T(x),y\rangle$).

Now someone has proved the below theorem. Is it true?


Let $X$ and $Y$ be two arbitrary Banach spaces. Then $F (X; Y^*)$(Finite rank operator) is dense in $B(X; Y^*)$ with respect to the weak* operator topology.


Let $T \in B(X; Y^*)$ and take a finite subset $F =\{ x_1,...,x_n\}$ of X. Assume that $x_1,...,x_m$ are linearly independent for all $m\leq n$. By the Hahn Banach theorem, for each $j\in \{1,2,...,m\}$ there is $f_j\in X^*$ such that $f_j(x_j) = 1$ and $f_j(x_i) = 0$ for all $i\in\{1,2,...,m\}-\{j\}$.For each $j\in\{1,...,m\}$ define $T_j\in B(X; Y^*)$ by $T_j(x) = f_j(x)T(x_j)$.Then $$T_j(x_j) = T(x_j); T_j(x_i) = 0\ \ (i, j\in\{1,...,m\}, i\neq j)$$ Now define $T_F = T_1 +...+ T_m$. It can be easily seen that $$T_F(x_i) = T(x_i)\ \ (i\in\{1,...,n\})$$ So $T_F = T$ on the span of $F$ and $\operatorname{rank}(T_F)\leq \dim F$. Now it is obvious that the net $(T_F)_{F\in F(X)}$$\big(F(X)=$ all finite subset of $X\big)$ converges to $T$ in the weak* operator topology, as desired.


If it is true then my question is this that why we can't say $T_F\rightarrow T$ in strong operator topology($T_F(x)\rightarrow T(x) \ \ \forall x\in X$)?

| cite | improve this question | | | | |
  • 1
    $\begingroup$ In the first line under "Proof" I suspect you meant to say "Assume $x_1,\ldots,x_m$ are linearly independent ($m \le n$)." $\endgroup$ – hardmath Feb 26 '14 at 11:27
  • $\begingroup$ We can assume that they are linearly independent.I think there is no any problem there. $\endgroup$ – Hamid Shafie Asl Feb 26 '14 at 15:25
  • 1
    $\begingroup$ I was trying to address two matters. One was the otherwise undefined introduction of $m$ (except to say some $m \le n$). The other is to address linear independence. Recall that $F$ is a typical finite subset of $X$. Generally such sets are not always linearly independent. Moreover the plan of the proof is to consider a net $\mathscr{F}$ upon all finite sets. For both these reasons it makes to distinguish a (maximal) linearly independent subset $\{x_1,\ldots,x_m\}$ of $F$, noting that the subscripts can be assumed like this because a set lacks a preferred ordering. $\endgroup$ – hardmath Feb 26 '14 at 18:49
  • $\begingroup$ A similar question is answered here math.stackexchange.com/questions/535645/… $\endgroup$ – Conifold Jun 13 '14 at 0:03

Indeed we also have $T_F \to T$ in the strong operator topology.

The strong operator topology is defined by the seminorms

$$p_F(S) = \sup \{\lVert S(x)\rVert : x \in F\},$$

where $F$ traverses the finite subsets of $X$. The construction immediately yields

$$p_F(T_F - T) = 0$$

for any linearly independent finite $F\subset X$, and it is straightforward to see $p_F(T_{F'} - T) = 0$ for any finite $F \subset X$ when $F' \subset F$ is a maximal linearly independent subset of $F$. That means every neighbourhood of $T$ in the strong operator topology contains operators of finite rank, so $F(X,Y^\ast)$ is also dense in the strong operator topology.

| cite | improve this answer | | | | |

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.