Let $X,Y$ be Banach spaces, $T:X\to Y$,unbounded linear operator.
How to prove that there is a natural number $n$,the set $\{x:\|Tx\|\le n\|x\|\}$ is dense in $X$?
Let $X,Y$ be Banach spaces, $T:X\to Y$,unbounded linear operator.
How to prove that there is a natural number $n$,the set $\{x:\|Tx\|\le n\|x\|\}$ is dense in $X$?
Copyright © 2021 JogjaFile Inc.
The claim is true (see the proof below using Baire's Category Theorem). But note that it is (without the axiom of choice) hard to come up with an unbounded operator $T$ that is defined on the whole Banach space $X$.
For $n\in\mathbb{N}$ define $$ A_{n}:=\left\{ x\in X \mid \left\Vert Tx\right\Vert \leq n\cdot\left\Vert x\right\Vert \right\} . $$ Note that $0\in A_{n}$ for all $n\in\mathbb{N}$ and that we have $rA_{n}\subset A_{n}$ for all $r>0$ and $n\in\mathbb{N}$.
Also note that we have $$ X=\bigcup_{n\in\mathbb{N}}A_{n}\subset\bigcup_{n\in\mathbb{N}}\overline{A_{n}}. $$ The Baire Category Theorems shows that there is $n_{0}\in\mathbb{N}$, $x_{0}\in X$ and $\varepsilon>0$ such that $$ B_{\varepsilon}\left(x_{0}\right)\subset\overline{A_{n_{0}}} $$ holds.
Choose $m\in\mathbb{N}$ with $$ \frac{2}{\varepsilon}\cdot\left[n_{0}\cdot\left(\left\Vert x_{0}\right\Vert +\varepsilon\right)+\left\Vert Tx_{0}\right\Vert \right]\leq m. $$
Now let $x\in X$ be arbitrary with $\frac{\varepsilon}{2}<\left\Vert x\right\Vert <\varepsilon$, so that $x_{0}+x\in B_{\varepsilon}\left(x_{0}\right)\subset\overline{A_{n_{0}}}$ holds. Thus, there is a sequence $\left(y_{n}\right)_{n\in\mathbb{N}}$ in $A_{n_{0}}$ such that $y_{n}\rightarrow x_{0}+x$ holds. This yields $y_{n}-x_{0}\rightarrow x$ and $$ \begin{eqnarray*} \left\Vert T\left(y_{n}-x_{0}\right)\right\Vert & \leq & \left\Vert Ty_{n}\right\Vert +\left\Vert Tx_{0}\right\Vert \\ & \leq & n_{0}\cdot\left\Vert y_{n}\right\Vert +\left\Vert Tx_{0}\right\Vert \\ & = & \underbrace{\frac{n_{0}\cdot\left\Vert y_{n}\right\Vert +\left\Vert Tx_{0}\right\Vert }{\left\Vert y_{n}-x_{0}\right\Vert }}_{=:\alpha_{n}}\cdot\left\Vert y_{n}-x_{0}\right\Vert .\qquad\left(\ast\right) \end{eqnarray*} $$ Now note that $\frac{\varepsilon}{2}<\left\Vert x\right\Vert <\varepsilon$ yields $$ \begin{eqnarray*} \alpha_{n}\xrightarrow[n\rightarrow\infty]{}\frac{n_{0}\cdot\left\Vert x_{0}+x\right\Vert +\left\Vert Tx_{0}\right\Vert }{\left\Vert x\right\Vert } & < & \frac{2}{\varepsilon}\cdot\left[n_{0}\cdot\left\Vert x_{0}+x\right\Vert +\left\Vert Tx_{0}\right\Vert \right]\\ & \leq & \frac{2}{\varepsilon}\cdot\left[n_{0}\cdot\left(\left\Vert x_{0}\right\Vert +\left\Vert x\right\Vert \right)+\left\Vert Tx_{0}\right\Vert \right]\\ & < & \frac{2}{\varepsilon}\cdot\left[n_{0}\cdot\left(\left\Vert x_{0}\right\Vert +\varepsilon\right)+\left\Vert Tx_{0}\right\Vert \right]\\ & \leq & m. \end{eqnarray*} $$ We thus get $\alpha_{n}<m$ for $n$ large enough. The estimate marked with $\left(\ast\right)$ above thus shows $y_{n}-x_{0}\in A_{m}$ for $n$ large enough and thus $x=\lim_{n}\left(y_{n}-x_{0}\right)\in\overline{A_{m}}$.
We have thus established $$ B_{\varepsilon}\left(0\right)\setminus\overline{B_{\varepsilon/2}}\left(0\right)\subset\overline{A_{m}}. $$ This implies $$ B_{r\varepsilon}\left(0\right)\setminus\overline{B_{r\varepsilon/2}}\left(0\right)\subset r\overline{A_{m}}=\overline{rA_{m}}\subset\overline{A_{m}} $$ for all $r>0$. But $$ X\setminus\left\{ 0\right\} =\bigcup_{r>0}\left[ B_{r\varepsilon}\left(0\right)\setminus\overline{B_{r\varepsilon/2}}\left(0\right)\right], $$ so that $X\subset\overline{A_{m}}$ follows. This completes the proof.