6

Is there a finite-dimensional vector space whose dimension cannot be found? Assume, we have somehow constructed a vector space whose dimension is finite, but yet unknown. Is there always an algorithm to find it?

Take the example by Noam Elkies from the comments: let $V$ be the vector subspace of $\mathbb{R}$ over $\mathbb{Q}$ spanned by $\lbrace 1,e,\pi \rbrace$. Should we believe that $\dim(V)$ is either $2$ or $3$, even if we don't know which?

flag
10 
 Consider the vector space whose dimension is $1$ if a particular Turing machine halts on a particular input and $0$ otherwise. – Qiaochu Yuan Oct 3 2011 at 18:56
41 
 Let $V$ be the span of $\lbrace 1, e, \pi \rbrace$ as a vector space over ${\bf Q}$. Then $\dim(V)$ is finite, and surely equal to $3$, but nobody knows how to prove it. – Noam D. Elkies Oct 3 2011 at 19:28
6 
 From a constructivist point of view, this is a very 'real' question. Noam Elkies' answer is quite a good one. One can expand on that pattern to give quite substantial answers to this question. Vote to reopen. – Jacques Carette Oct 3 2011 at 19:33
8 
 @Reimundo Heluani: you make it sound as if editing and closing where mutually exclusive. In some sense, the opposite is the case. A main point of closing a vague question is to give time for editing, while (temporarily) holding back answers that then, after a clarifiying edit, might seemm off-topic or besides the point. For a more authorative account (not for this specific question but in general) please see François's contribution, in particular 3. in his numbered list, in the following recent meta thread meta.mathoverflow.net/discussion/1156/… – quid Oct 3 2011 at 23:14
8 
 on META: meta.mathoverflow.net/discussion/1158/… – Will Jagy Oct 4 2011 at 0:36
show 11 more comments

closed as not a real question by Igor Rivin, Qiaochu Yuan, Steve Huntsman, Ricky Demer, quid Oct 3 2011 at 19:13

Browse other questions tagged or ask your own question.