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 am reading de la Harpe's book "Topics in Geometric Group Theory". On page 145, there is a theorem: Let $V$ be a complete $n$-Riemannian manifold with sectional curvature satisfying $K\ge 0$. Then there exist a finite normal subgroup $F_1$ of the fundamental group $\pi_1(V)$ and a short exact sequence of the form: $$ 1 \to \mathbb Z^k \to \pi_1(V)/F_1 \to F_2 \to 1 $$ where $k\le n$ and where $F_2$ is a finite group.

I have two questions:

1) why the author put $F_1$ in such a special position, why not absorb $F_1$ into $F_2$, since they are basically all finite group.

2) In other papers/books, peopole called a discrete, torsion free and cocompact subgroup of $O(n)\ltimes \mathbb R^n $ A Bieberbach group. So why "torsion free" is important?

Thank you!

share|cite|improve this question
In 1) you can indeed absorb $F_1$ into $F_2$, see Theorem 2.1 in Wilking's preprint For 2) the torsion-free case is easier and cleaner, and also these are the groups of compact flat manifolds, which matter for geometers. – Igor Belegradek Sep 5 '12 at 14:43
@Igor, Thanks. Is it true that the groups mentioned in Wilking's paper Thm2.1 is $\mathbb Z^n\ltimes F$ where $F$ is a finite group. The difference between the Flat Bieberbach group and the one in Wilking's list is for Bieberbach group $F$ can be considered as a subgroup of $O(n)$. However for non-negatively curved case, it is a subgroup of $GL(n, \mathbb Z)$? – John B Sep 5 '12 at 18:22
The author's name is "de la Harpe". – Igor Rivin Sep 5 '12 at 18:39
@MG: The groups in Thm 2.1 need not be semidirect products, eg the fundamental group of a flat manifold is torsion-free, so it cannot contain nontrivial finite subgroups. Your other question seems misstated for any finite subgroup of $GL_n(\mathbb R)$ preserves an inner product, so it lies in a conjugate of $O(n)$. In any case a finitely generated group is crystallographic iff it is virtually abelian and has no normal finite subgroups. Thus torsion-free crystallographic groups (also known as Bieberbach groups) are precisely the torsion-free virtually-$\mathbb Z^n$, see rmk 2.5(2) in Wilking. – Igor Belegradek Sep 5 '12 at 18:47
That you can absorb $F_1$ into $F_2$ follows from the fact that polycyclic-by-finite groups are residually finite (I think this is due to Malcev). Actually, here you reduce to the finite-by-abelian case, which easily boils down to showing that every f.g. 2-nilpotent group is residually finite; this is a simple exercise. – YCor Sep 5 '12 at 22:36
up vote 3 down vote accepted

There is a ``quasi-isometric rigidity'' context in which a finite kernel arises naturally. I am not sure how general this is so let me state it in the narrower context of constant sectional curvature equal to zero.

Theorem (quasi-isometric extension of the Bieberbach Theorem): If $\Gamma$ is an abstract group quasi-isometric to Euclidean $n$-space then there exists a finite normal subgroup $F_1$ of $\Gamma$ such that $\Gamma / F_1$ is a Bieberbach group (of the general kind, allowing torsion), and there is a short exact sequence of the form $1 \to \mathbb{Z}^n \to \Gamma / F_1 \to F_2 \to 1$ where $F_2$ is a finite group. So the group $\Gamma/F_1$ is the orbifold fundamental group of a compact $n$-dimensional Euclidean orbifold.

This is a consequence of Gromov's polynomial growth theorem. I don't know whether this is what de la Harpe had in mind, but maybe.

share|cite|improve this answer
A finite kernel also arises if you characterize discrete (not necessarily cocompact) groups of isometries of Euclidean spaces. – YCor Sep 5 '12 at 22:39

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.