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

$$ U \; = \; \left( \begin{array}{cc} 0 & 1 \\\ 1 & 0 \end{array} \right) , $$

Given a real oriented vector space $V$ with inner product, form Lorentzian $L = V \oplus U.$ Elements are of the form $ (v; x,y). $ The norm on $L$ is given by $$ (v; m,n)^2 = v^2 + 2 xy.$$ We infer the inner product $$ (v_1; x_1,y_1) \cdot (v_2; x_2,y_2) = v_1 \cdot v_2 + x_1 y_2 + x_2 y_1.$$

Given any $ l \in L,$ a null vector, we have $l \cdot l = 0,$ and so $l \in l^\perp.$ Furthermore, if $k \in l^\perp$ as well, then $ (k + l)^2 = k^2.$ As a result, we may form the one-dimensional space $\langle l \rangle$ spanned by $l$ itself, then form another space with norm, $$ E(l) = l^\perp / \langle l \rangle.$$ We have that $E\left((0;0,1) \right) = V,$ and $E(l)$ is positive definite.

Question: is there a reasonably consistent way to assign an orientation to all the $E(l)?$

Motivation: the Leech lattice is chiral. That is, there is no automorph of the Leech lattice with negative determinant. All the even lattices that Pete Clark and I found are achiral, they all possess improper automorphs. So, not only are they in genera of class number one, they are in genera of proper class number one. I am trying, with a good deal of frustration, to decide whether Conway's argument, as I report in A Priori proof that Covering Radius strictly less than $\sqrt 2$ implies class number one really imples proper class number one, or is it just luck? After some email with Daniel Allcock, this question is a beginning. Daniel emphasizes that orientation of a lattice is an orientation of the surrounding real vector space. Conway's proof is that (in my case) every (primitive) null vector is equivalent by a sequence of Lorentz reflections in roots, which ought certainly to be said to reverse orientation on $L.$ But the $E(l)$ construction does not seem to care about that, it does not know how we got $l.$ At the end, and I may need several questions to get through this, does Conway's argument actually show proper class number one?

Very confused.

share|cite|improve this question
up vote 1 down vote accepted

This particular question (``consistently orienting the reductions") and all those arising from your above link are interesting.

However, the answer to this question seems `easy' -- and i'm wondering if i'm not missing something.

First, let us fix an orientation on the total lorentz space $V':=V \oplus U$ and let $h$ denote our lorentz metric $h(v,u_1, u_2)=v^2+2u_1v_1$.

For a nonzero null vector $n \in V'$, your question is really to find a smooth co-orientation of the subspaces $n^\perp$, i.e. we want a smooth family of nonzero oriented transversals to $n^\perp$ (depending smoothly on the nonzero null vector $n$).

Now for something that's almost silly. In analogy with almost-complex structures in symplectic geometry, consider those linear involutions $j$ of $V'$ (so $j^2=+id$) with the property that $h(x,jx)>0$ for all nonzero $x\in V'$ (we might say that $j$ `tames' $h$). Such involutions do exist, e.g. all $j_\lambda:=\begin{pmatrix} \lambda & 0 \\ 0 & -\lambda^{-1} \end{pmatrix}$ for $\lambda \neq 0$ are involutions taming $U$ with respect to the standard null basis.

The totality of such $j$'s (I wouldn't call them almost-complex structures ...) taming $h$ (or additionally acting as $h$-isometries) form a contractible symmetric space (by the same `fibration' argument that Gromov uses in his 1985 paper on pseudo-holomorphic curves).

As I see it, the existence of $j$'s taming your lorentz metric $h$ is exactly what is needed to co-orient the subspaces $n^\perp$, since indeed we have the splitting $V'=n^\perp \oplus \mathbb{R} jn$.

This argument shows that we could not expect to co-orient the reductions $E(\mathbb{R}n)$ as they vary over the unoriented null lines $\mathbb{R}n$.

share|cite|improve this answer
Thank you. Pete Clark just submitted our joint paper with the Lorentzian stuff as the last two sections. I do not believe he included the orientation idea at all. The way it all turned out, the only example where there was any doubt possible was $2 x^2 + x y + 2 y^2,$ which is "ambiguous" by inspection. – Will Jagy Dec 31 '13 at 21:39
@WillJagy: Your related posts (especially euclidean implies class number one) have been very interesting and helpful for me. I look forward to reading your paper. – J. Martel Jan 1 '14 at 5:59
The version of the 13th of November is item number 28 at There was at least one more version, Pete submitted the thing on 29th November. If you email me I can send you the version that was actually submitted. Meanwhile, I think the technique used for class number one probably has other uses, apart from integral lattice automorphism groups and genus size. – Will Jagy Jan 1 '14 at 6:14

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.