4
$\begingroup$

Given an integral domain $R$, the Steinberg group $St_n(R)$ is the group given by generators
$e_{pq}(\lambda) := \mathbf{1} + a_{pq}(\lambda)$, $p\neq q$, $1\leq p,q \leq n$

Subject to the relations
$$\begin{align} e_{ij}(\lambda) e_{ij}(\mu) &= e_{ij}(\lambda+\mu) \\ \left[ e_{ij}(\lambda),e_{jk}(\mu) \right] &= e_{ik}(\lambda \mu) && \mbox{for } i \neq k\\ \left[ e_{ij}(\lambda),e_{kl}(\mu) \right] &= \mathbf{1} && \mbox{for } i \neq l, j \neq k\\ \end{align}$$

The Steinberg group is the universal central extension of the special linear group over $R$; $Sl_n(R)$.

Is there a description of the Steinberg group $St_n(Z)$, the special linear group over the integers as a lattice in some lie group, and some covering map realizing the universal central extension of $Sln(R)$ ( real coefficients), which restricts to the integral universal central extension of $Sln(Z)$ given by the Steinberg group ?

$\endgroup$

2 Answers 2

6
$\begingroup$

For $n\ge 3$ the Steinberg group $\text{St}_n(\mathbf{Z})$ is a lattice in the universal covering of $\text{SL}_n(\mathbf{R})$ (which has fundamental group the cyclic group of order 2).

To see that the inverse image of $\text{St}_n(\mathbf{Z})$ in the universal covering of $\text{SL}_n(\mathbf{R})$ is indeed a non trivial central extension, essentially amounts to check that the image of $K_2(\mathbf{Z})$ in the topological $K_2$ or the reals is injective (but I guess that a suitable direct argument, for instance using an element of order 2, can work as well).

$\endgroup$
1
  • $\begingroup$ I mean "of the reals", sorry. $\endgroup$
    – YCor
    Aug 21, 2012 at 22:09
1
$\begingroup$

EDIT: This started to be an answer, but my recollections were inaccurate. It's essential here to distinguish carefully between what happens when $n=2$ (where life is much more complicated) and when $n \geq 3$ (where the situation stabilizes). The definition of the Steinberg group differs in these situations.

$\endgroup$
2
  • 2
    $\begingroup$ @Jim: I respectfully disagree: $SL_n(\mathbb{R})$ is NOT simply connected for $n\geq 3$ (it is homotopy equivalent to $SO(n)$), so its universal cover is precisely of degree 2, and therefore the answer for the Steinberg group seems to be affirmative. $\endgroup$ Aug 21, 2012 at 21:54
  • $\begingroup$ @Alain: On further thought, I see that I was too hasty in my "answer", so I've edited most of it out. Sorry for the confusion. $\endgroup$ Aug 22, 2012 at 0:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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