Without (at least a weak form of) choice, we cannot prove the existence of a basis for $\mathbf{R}$ over $\mathbf{Q}$. But is there some vector space $V\supset \mathbf{R}$ that we can prove has a basis, without choice?

Note that it is not true in general that if $V$ has a basis and $U\subseteq V$ then $U$ has a basis; see https://mathoverflow.net/questions/80765/if-v-is-a-vector-space-with-a-basis-w-subseteq-v-has-to-have-a-basis-too

  • 1
    $\begingroup$ @Asaf Karagila I think this is not a duplicate of math.stackexchange.com/questions/257325/… because that question gets us "There are models where R has no basis over Q" but that does not get us "There are models where no superspace of R has no basis over Q"; in particular, I do not see why a superspace of R having a basis implies R has a basis. $\endgroup$ – alphacapture Sep 30 '17 at 23:56
  • $\begingroup$ You are absolutely right. I misread the question. $\endgroup$ – Asaf Karagila Oct 1 '17 at 5:58

No. If there existed such a basis, we could define the following linear function $f:V\to\mathbf{Q}:$

Write $1\in\mathbf{R}$ as a linear combination of the basis vectors. Let $v$ be one of those basis vectors. Send $v$ to 1 and the rest to 0.

But now the restriction of this to $\mathbf{R}$ would be a linear function from $\mathbf{R}$ to $\mathbf{Q}$ other than the 0 function, which does not exist without some amount of choice.


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.