Exercise :

Show that if $X$ is a finite-dimensional normed space and $Y$ is a normed space, then every linear operator $T:X \to Y$ is bounded.

Attempt :

Since $X$ is finite-dimensional, say $\dim(X)=n$, there exists a basis $\{e1,e2,...,en\}$ of $X$ such that every element $x\in X$ can be written uniquely in the form:

$$\begin{align} \quad x = a_1e_1 + a_2e_2 + ... + a_ne_n \end{align}$$

where $a_1,a_2,\dots,a_n \in \mathbb R$.

Now, $\forall \; x \in X$, it is :

$$\begin{align} \quad \| T(x) \|_Y &= \| T (a_1e_1 + a_2e_2 + ... + a_ne_n) \|_Y \\ &= \| a_1 T(e_1) + a_2 T(e_2) + ... + a_n T(e_n) \|_Y \\ & \leq \sum_{k=1}^{n} |a_k| \| T(e_k) \|_Y \end{align}$$

Using the Cauchy-Schwarz inequality, we yield :

$$\begin{align} \quad \| T(x) \| & \leq \left ( \sum_{k=1}^{n} |a_k|^2 \right )^{1/2} \left ( \sum_{k=1}^{n} \| T(e_k) \|_Y^2 \right )^{1/2} \\ & \leq \left ( \sum_{k=1}^{n} |a_k|^2 \right )^{1/2} \cdot M \end{align}$$

But regarding equivalence of norm in correlation to finite-dimensional spaces, we have that :

$$\begin{align} \quad \| T(x) \|_Y & \leq M \| x \|_* \end{align}$$

Then, $\exists c_1,c_2 \in \mathbb R^+ : \forall x \in X$ it is :

$$\begin{align} \quad c_1 \| x \|_X \leq \| x \|_* \leq c_2 \| x \|_X \end{align}$$

Thus $\forall x \in X$ it is :

$$\begin{align} \quad \| T(x) \| & \leq c_1M \| x \|_X \end{align}$$

which tells us that $T$ is bounded.

Question : It seemed like a rather hard exercise to me so I am not sure if my proof/approach is definitely correct or rigorous enough. Any insight will be very helpful !

  • $\begingroup$ from the inequality $\| T(x) \|_Y \leq \sum_{k=1}^{n} |a_k| \| T(e_k) \|_Y$ the proof is almost done because clearly $\|T(x)\|_Y<\infty$, so you have a linear operator that map bounded sets to bounded sets, what is the definition of bounded linear operator. $\endgroup$ – Masacroso Nov 3 '18 at 16:22
  • $\begingroup$ @masacroso True ! I just tried to finish it to have the standard bounded form. $\endgroup$ – Rebellos Nov 3 '18 at 16:32
  • $\begingroup$ @Masacroso: "bounded" is not a magical word. The notion of "bounded" depends on a metric. It is not obvious that different metrics will give you the same bounded sets. It does work for metrics given by norms on a finite-dimensional space, because of the nontrivial fact that all norms are equivalent, as used by the OP. $\endgroup$ – Martin Argerami Nov 4 '18 at 19:14

Your proof is both correct and rigorous. The only change I would suggest is that, instead of using Cauchy-Schwarz, you can take $M=\max\{\|Te_k\|:\ k=1,\ldots,n\}$ and then $$ \sum_{k=1}^{n} |a_k| \| T(e_k) \|_Y \leq M\,\sum_{k=1}^n|a_k|, $$ and you can use that norm to compare with the original.

  • $\begingroup$ Thanks a lot for your input ! This is a pretty nice suggestion and it simplifies the proof a bit (makes it a lot shorter). Would it be more "elegant" to approach it like that ? $\endgroup$ – Rebellos Nov 4 '18 at 21:42
  • $\begingroup$ There's no canonical answer. But I'm all for typing less. $\endgroup$ – Martin Argerami Nov 4 '18 at 22:15

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.