Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

In this answer, the OP gave a proof to the proposition that if a linear functional $f$ is not bounded, then it has a non-closed kernel. However, he only showed that if $f$ is not bounded, then a unbounded sequence exists in the kernel, which doesn't necessarily prove that the kernel is non-closed, since closed sets obviously can have unbounded sequences. Am I wrong? If that proof was wrong, then what's the correct proof of this proposition?

share|cite|improve this question
I think the answer might have confused the definitions of closure and compactness – Omnomnomnom Dec 11 '13 at 21:18
up vote 5 down vote accepted

Assuming we are working with normed spaces here, pick a fixed vector $v$ with $f(v)=1$. If $f$ is not bounded, there is a sequence $(x_n)$ with $x_n\to0$ and $f(x_n)=1$. Thus $v-x_n\in\operatorname{ker} f$. But $v-x_n\to v\notin\operatorname{ker}f$, so $\operatorname{ker}f$ is not closed.

Addendum: The answer is already accepted, but for future reference, here is how to generalize the above to a topological vector space. It requires a little lemma, which is not hard to prove (I leave the proof to the reader): If $f$ is an unbounded linear functional, then every neighbourhood of the origin contains some $x$ with $f(x)=1$. Assuming this, as in the above case start with a fixed vector $v$ with $f(v)=1$. If $U$ is a neighbourhood of $v$, then $v-U$ is a neighbourhood of $0$, so it contains some $x$ with $f(x)=1$. Thus $v-x\in U\cap\operatorname{ker}f$, and we're done.

share|cite|improve this answer
Thanks. And why exactly we can find such sequence $x_n$? – Voldemort Dec 11 '13 at 21:53
Oh, I get it. It's because according to one of the definitions of unbounded linear operator, there exists $y_n$, with $\|y_n\|=1$ but $|f(y_n)|\to\infty$, now let $x_n=y_n/|f(y_n)|$ and we get what we want. – Voldemort Dec 11 '13 at 23:13

For a linear map between normed spaces, boundedness is equivalent to continuity at $0$, which is equivalent to continuity everywhere.

If $f\colon V\to\mathbb{R}$ (or $\mathbb{C}$) is a linear map and $f$ is continuous (that is, bounded), then $\ker f$ is the inverse image of $\{0\}$, so it's closed.

If $\ker f$ is closed, then $f$ induces a map from $V/\ker f$ (which is a normed space) to $\mathbb{R}$. If $f$ is nonconstant (otherwise the result is trivial), then the induced map is an isomorphism and it is continuous because such is any linear map between finite dimensional normed spaces. Therefore $f$ is the composition of a continuous linear map with the projection $V\to V/\ker f$, which is continuous by definition.

share|cite|improve this answer
Why is $V/\ker f$ a finite dimensional space? – Voldemort Dec 11 '13 at 22:06
@Voldemort Because it's isomorphic to $\mathbb{R}$. – egreg Dec 11 '13 at 22:07
Btw, this answer was also given to the original question. – Carsten S Dec 12 '13 at 8:50
@egreg why is it isomorphic to $\mathbb R$ and not $\operatorname{Im}f$? – Arrow Jul 11 at 13:45
@Arrow The image of $f$ is $\mathbb{R}$, because $f$ is unbounded, so the image cannot be $\{0\}$. – egreg Jul 11 at 13:54

A correct proof is given by Beni Bogosel in his blog.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.