Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I got this questions on one of my example sheets in a first year algebra course:

"Let $B:W×V→K$ be a bilinear map, where $V,W$ are vector spaces over a a field $K$. Let $U$ be a subspace of $V$. If we denote by $U^{\perp}= \{w \in W|B(w,u)=0,\forall u∈U \},$ prove that $dim(U)+dim(U^{\perp})≥dim(W)$.

I managed to come up with a proof in the case when $V,W$ are finite dimensional by considering a basis of $V$ which contains a basis of $U$, a basis of $W$ and then considering the coordinate matrix of $B$.

I can't think of an argument which works for the infinite dimensional case. Could somebody suggest me something? Thank you

share|cite|improve this question

This is false in the sense of cardinal arithmetic if the spaces are allowed to be infinite dimensional. Take $U=V$ any infinite dimensional space, $W=U^*$, and $B$ the natural pairing $(w,u)\mapsto w(u)$. Then $\dim U^\perp=0$ but $\dim U<\dim U^*=\dim W$.

For a concrete example take $W=K[[X]]$ (uncountable dimensional) and $U=K[X]$ (countable dimensional) with the "standard" pairing between these two $K$ vector spaces.

share|cite|improve this answer
Thank you! Indeed I was wondering if it's true in cardinal arithmetic. – turcas Nov 11 '12 at 9:08

The dimension of $V$ appears to be irrelevant, as we can replace it with $U$. So the infinite-dimensional case is really when $\dim \, (W) = \infty$, and so we need to show that one of the quantities on the left-hand side is infinite. Suppose that $\dim \, U$ is finite. What can you conclude about $\dim \, U^\perp$?

Hint: Since $\dim \, U$ is finite, we can find a basis $e_1, \cdots, e_n$ and construct linear functionals $L_i : W \to K$ defined by $L_i(w) = B(w,e_i)$. Clearly, $U^\perp = \bigcap_{i=1}^n \ker L_i$. For each $L_i$, what are the possible dimensions of $W / \ker L_i$ -- the subspace of $W$ of all vectors not in the kernel of $L_i$? What can you conclude about the dimension of $W/ \bigcap_{i=1}^n \ker L_i$? When $\dim \, W = \infty$, what does that tell you about the dimension of $\bigcap_{i=1}^n \ker \, L_i$, and hence the dimension of $U^\perp$?

share|cite|improve this answer
You're assuming that the question is only about infinite or not; I'd interpreted it as referring to cardinal arithmetic. – joriki Nov 11 '12 at 8:35
I don't think they teach cardinal arithmetic in first year algebra classes. In my experience, these sorts of problems were always about finite vs infinite, but that's certainly something to consider. – Isaac Solomon Nov 11 '12 at 8:48
See the OP's comment under Marc's answer. – joriki Nov 11 '12 at 10:08

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.