Question: For every Banach space $X$ and its subspace $Y$, is there a complemented subspace $Z$ in $X$ such that $Y \subset Z \subset X $ and $\operatorname{card}(Y)=\operatorname{card}(Z)$ i.e., $Y$ and $Z$ have the same cardinality?

The answer is yes if $\operatorname{card}(X)=\mathfrak c$ (i.e., Continuum), the proof is easy by taking a hyperplane $H$ such that $Y \subset H \subset X$, then $\operatorname{card}(Y)=\operatorname{card}(H)=\mathfrak c$.

But I don't know whether it holds for general spaces.

  • $\begingroup$ I assume by "subspace" you mean "closed subspace", right? $\endgroup$ – Theo Sep 12 '12 at 13:31
  • $\begingroup$ Yes, I mean "closed", and I has known the answer is no. See the reference : P.~Koszmider, \emph{A space $C(K)$ where all nontrivial complemented subspaces have big densities.} Studia Math. {\bf 168} (2005) 109--127 $\endgroup$ – Duanxu Dai Sep 13 '12 at 7:00
  • 1
    $\begingroup$ For those with institutional access: journals.impan.gov.pl/sm/Inf/168-2-2.html @Duanxu: would you be so kind as to post this comment into an answer box below, so that this question can be considered settled? $\endgroup$ – t.b. Sep 13 '12 at 7:35

The above paper says that there is a Banach space (of continuous functions on a totally disconnected compact Hausdorff space) of density κ bigger than the continuum which has no infinite-dimensional complemented subspaces of density continuum or smaller. In particular no separable infinite-dimensional subspace has a complemented superspace of density continuum or smaller. Hence, for the space C(K) constructed by P.~Koszmider, for each separable subspace Y of C(K), there exist no complemented subspace Z containing Y such that card(Z)=card(Y), in fact, such space Z has card lager than c.


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.