1
$\begingroup$

As in the title, my question is: What is the algebraic dual space of $\mathbb{R}^{\mathbb{N}}$ - the space of all $\mathbb{R}$-valued sequences. I think that it should be the direct sum $\bigoplus_{\mathbb{N}} \mathbb{R}$, via the natural duality, but I cannot proof it.

$\endgroup$
3
$\begingroup$

It's not the dual. Denote by $\def\<#1>{\left<#1\right>}\<,> \colon \mathbf R^{\mathbf N} \times (\mathbf R^{\mathbf N})^* \to \mathbf R$ the duality. We embed $\def\s{\mathbf R^{(\mathbf N)}}\s := \bigoplus_{\mathbf N}\mathbf R$ in $(\mathbf R^{\mathbf N})^*$ via $\def\p{\mathbf R^{\mathbf N}}$ $$ \<(a_n), (b_n)> := \sum_n a_n b_n, \quad a \in \p, b \in \s $$ We will show that this map is not onto. Let $c \subseteq \p$ denote the subspace of converging sequences, and let $\ell \in (\p)^*$ be a linear map such that $\<a,\ell> = \lim a_n$ for all $a \in c$. Suppose $\ell$ were represented by some $b \in \s$. Let $e^k \in c$ denote the sequence $(\delta_{kn})_n$, then $$ \ell(e^k) = \lim_n \delta_{kn} = 0 $$ Hence $$ b_k = \<b, e^k> = \ell(e^k) = 0$$ So $b =0$, but $\ell\ne 0$. That is $\ell \in (\p)^* \setminus \s$.

$\endgroup$
  • $\begingroup$ I do not believe that this is a counterexample. Why should it be possible to extend the limit operator $\ell$ to $\mathbb{R}^{\mathbb{N}}$? $\endgroup$ – Sebastian Sep 23 '15 at 13:30
  • 1
    $\begingroup$ By the axiom of choice. Let $B$ a basis of $c$. Extend it to a basis $B \uplus C$ of $\mathbf R^{\mathbf N}$. Define $\ell$ on $B \uplus C$ be $\ell(b) := \lim_n b_n$, $b \in B$ and $\ell(c) := 0$, $c \in C$, extended by linearity. $\endgroup$ – martini Sep 23 '15 at 13:32
  • $\begingroup$ Ok, you are right! Of course any subspace has an algebraic direct compliment. Thanks! $\endgroup$ – Sebastian Sep 23 '15 at 13:37

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.