Consider $\ell^\infty$ as a subspace of the Polish space $\mathbb{R}^\omega$. It is easy to check that $\ell^\infty$ is not Polish in the subspace topology, as it is countable union of the compact nowhere dense sets $\{f\in\mathbb{R}^\omega:\forall m(|f(m)|\leq n)\}$, and hence is not Baire.

Recall a topological group $(G,\tau_0)$ is Polishable if there is a Polish group topology $\tau_1$ on $G$ having the same Borel sets as $\tau_0$.

Is $\ell^\infty$, with the subspace topology above, Polishable? (I believe the answer is no.)

Note that it is easy to see, using Pettis' Theorem, that any such Polish topology would be a refinement of the subspace topology.

An elementary proof would be prefferred to one which shows that the Borel equivalence relation $E_1$ embeds into to the coset equivalence relation $\mathbb{R}^\omega/\ell^\infty$, and then cites Theorem 4.2 in Kechris-Louveau, The Classification of Hypersmooth Borel Equivalence Relations.


The answer is no.

Su Gao $\textit{Invariant Descriptive Set Theory}$, Lemma 9.3.3 has a direct proof.

As people usually do, denote the equivalence relation $\ell_\infty : = \mathbb{R}^\omega \backslash \ell_\infty$ (where of course the latter $\ell_\infty$ refers to the group).

$\ell_\infty$ is universal under $\leq_\text{B}$ for $K_\sigma$ equivalence relations. (Rosendal. See Theorem 8.4.2 of Gao.) $E_1$ is $K_\sigma$. Hence $E_1 \leq_\text{B} \ell_\infty$. If the group $\ell_\infty$ is Polishable, then the equivalence relation $\ell_\infty := \mathbb{R}^\omega \backslash \ell_\infty$ is a Borel orbit equivalence relation. It is well-known that $E_1$ is not Borel reducible to any orbit equivalence relation of a Borel group action (Kechris-Louveau; see Theorem 10.6.1 of Gao).

  • $\begingroup$ You can use markdown to get italics, bold and some other formatting. For example, instead of $\textit{Invariant Descriptive Set Theory}$ you can write *Invariant Descriptive Set Theory*. Compare $\textit{Invariant Descriptive Set Theory}$ and Invariant Descriptive Set Theory. $\endgroup$ – Martin Sleziak Jun 12 '14 at 6:05
  • $\begingroup$ @MartinSleziak Thanks. That would be easier. I will remember this tip in the future. $\endgroup$ – William Jun 12 '14 at 7:46

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.