2
$\begingroup$

When studying the Lorentz group $O(1,3)$, one can decompose it into four parts... physicist usually called these

  • Proper-orthochronuos $\mathscr{L}^{\uparrow}_+$,
  • Proper-asynchronous $\mathscr{L}^{\downarrow}_+$,
  • Improper-orthochronuos $\mathscr{L}^{\uparrow}_-$,
  • Imroper-asynchronous $\mathscr{L}^{\downarrow}_-$,

depending on the determinant and the sign of the time-time element of the matrix representation.

Is it possible to define a similar decomposition for $O(n,p)$ groups? How could it be done?

Thank you in advance.

$\endgroup$
2
  • 2
    $\begingroup$ On page 20 of Lawson & Michelsohn it says "It is a classical fact (cf. Helgason [1]) that $SO(n)$ is connected and that $SO(r,s)$, for $r,s \geq 1$, has exactly two connected components. It is also a classical fact that $\pi_1SO(n) \cong \mathbb{Z}_2$ for $n \geq 3$ and $\pi_1SO(r,s) \cong \pi_1SO(r) \times \pi_1SO(s)$ for all $r,s$. Hence, $\pi_1SO^0(1,r) = \pi_1SO^0(r,1) = \mathbb{Z}_2$ and $\pi_1SO^0(r,s) = \mathbb{Z}_2 \times \mathbb{Z}_2$ for all $r,s \geq 3$. (Here $SO^0(r,s)$ denotes the connected component of the identity.)". I didn't see a proof in Helgason but I didn't look too hard. $\endgroup$ Commented Jul 30, 2012 at 20:22
  • $\begingroup$ @paul-reynolds Thank you for the comment, it's a very sketchy proof of the statement. I'll look Lawson & Michelsohn's and Nakahara's books for more details. $\endgroup$
    – Dox
    Commented Jul 30, 2012 at 21:37

1 Answer 1

6
$\begingroup$

Provided that $n,p>0$, $O(n,p)$ has four connected components as well. There are many ways to see this. $O(n,p)$ is a matrix subgroup of the general linear group of $\mathbb{R}^{n+p}$: $$ O(n,p) = \lbrace a \in \operatorname{GL}(n+p,\mathbb{R}) \mid a^T \eta a = \eta\rbrace$$ where $$\eta = \begin{pmatrix} -I_n & 0 \cr 0 & I_p \end{pmatrix}$$ with $I_k$ the $k\times k$ identity matrix.

It follows from the defining equation $a^T \eta a = \eta$ by taking determinants that $\det a = \pm 1$. We let $SO(n,p)$ denote the normal subgroup consisting of $a \in O(n,p)$ with $\det a = 1$. Then if $a_- \in O(n,p)$ is any matrix with $\det a_- = -1$, then $$O(n,p) = SO(n,p) \sqcup SO(n,p)\cdot a_-$$ where the union is disjoint. This shows that $O(n,p)$ is diffeomorphic to the disjoint union of two copies of $SO(n,p)$ and it remains to study the connectedness properties of $SO(n,p)$.

$SO(n,p)$ is homotopy equivalent to its maximal compact subgroup, which by polar decomposition is $$S(O(n)\times O\(p\)) = \lbrace (a_1,a_2) \in O(n)\times O\(p\) \mid \det a_1 \det a_2 = 1 \rbrace$$

This can be shown to have two connected components, by a very similar argument to the one at the start of this answer. Each such component is diffeomorphic to $SO(n) \times SO\(p\)$. Now use that $SO(m)$ is connected to conclude that $SO(n) \times SO\(p\)$ is as well.

This gives you the four connected components of $O(n,p)$.

Basic facts like this about classical Lie groups can be found in Wulf Rossmann's book "Lie groups, an introduction through linear groups."

$\endgroup$
1
  • $\begingroup$ Thank you Jose. A very complete answer! I'll check the recommended reference. CHeers $\endgroup$
    – Dox
    Commented Jul 30, 2012 at 21:39

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .