MathOverflow is a question and answer site for professional mathematicians. 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 hope my question is not too trivial. Let's suppose we have a vector space $V$ with a unimodular quadratic form $q$ of signature $(m,n)$.

My question is: which is the maximum dimension of an istropic subspace of $V$?

I would say it is $min\{m,n\}$. This is why. Suppose $min\{m,n\}=m$, if $min\{m,n\}=n$ the proof is the same. Given $U$, a positive subspace of dimension $m$, i choose $u_1,\cdots, u_m$ an ortonormal basis of $U$. Then i consider $U^\perp$ which is negative and i choose $w_1,\cdots,w_n$ an ortonormal basis. So the vectors $\{u_1+w_1,u_2+w_2,\cdots,u_m+w_m\}$ span an isotropic subspace of dimension $m$.

But could there be an isotropic subspace of dimension more than $min\{m,n\}$?

share|cite|improve this question
You could consider the projection map onto the positive definite subspace? – Tom Lovering Jun 26 '13 at 11:25
sorry but i don't follow you – michael waltz Jun 26 '13 at 11:36
@Tom Lovering: There is no such intrinsic thing as "the" positive-definite subspace (of maximal dimension, I suppose) in the indefinite case, so your suggestion is unclear. – user61789 Jun 26 '13 at 12:52
@michael: Please clarify in the question if you are working with vector spaces over $\mathbf{R}$ or not. It is a bit confusing, because you speak of $q$ being "unimodular", which suggests you have in mind a rational and even integral structure on $V$, raising the possibility that you are working over $\mathbf{Q}$ (briefly, why do you say "unimodular", and what do you mean by that when working over $\mathbf{R}$ if that is the intent?). – user61789 Jun 26 '13 at 12:54
Of course I meant $U$. So I was suggesting if you take the projection map $V \rightarrow U$, its kernel is negative definite, so in particular intersects any isotropic subspace trivially, so under this map, isotropic subspaces are embedded as subspaces of U. – Tom Lovering Jun 27 '13 at 11:21
up vote 2 down vote accepted

This is the Witt index of the quadratic form. I'm assuming you're working over $\bf{R}$, in which case it is indeed $\min\{m,n\}$ (use Witt cancellation). It can be smaller over other fields (like $\bf{Q}$); for instance, you could have an indefinite quaternary form which is anisotropic. For more details see Chapter 1 of Milnor and Husemoller's "Symmetric Bilinear Forms" or Lam's "Introduction to Quadratic Forms over Fields".

share|cite|improve this answer

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.