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

Let $R=\mathbb{Q}[e^{2\pi i /3}]$. Does $H_3(GL(R))$ have $\mathbb{Z}$-rank $1$?

If so, what is the index of the map: $$ \mathbb{Z}\cong K_3(R)/{\rm Torsion} \to H_3(GL(R))/{\rm Torsion}\cong \mathbb{Z}? $$

share|cite|improve this question
No, it doesn't have $Z$-rank 1. If you replaced $R$ with $Z[e2πi/3]$, it would. Or if you replaced $GL$ by $SL$. – Ben Wieland Aug 9 '11 at 1:12
Let me add a proof to Ben's comment: The the multiplicative group $R^\times$ of a number field is isomorphic to the product of the unit group of its ring of integers and a free abelian group $A$ of countably infinite rank. Since $GL(R) \cong SL(R) \times R^\times$ it follows from Kunneth formula that $H^3(GL(R);\mathbb{Z})$ contains $H^3(A;\mathbb{Z})$ as a direct summand. But $H^3(A;\mathbb{Z})$ has a subgroup isomorphic to $\Lambda^3(A)$ (see [Brown, Cohomology of Groups, Theorem 6.4]), which has clearly infinite rank. – Ralph Aug 9 '11 at 9:41
up vote 6 down vote accepted

Let $F = \mathbb{Q}(\zeta_3)$ and $R = \mathbb{Z}[\zeta_3]$ with a third root of unity $\zeta_3$. Then the Hurewicz homomorphism $$h_3: K_3(R) \to H_3(GL(R); \mathbb{Z})$$ induces an isomorphism on the torsion-quotients und the index of the map is $1$. This can be seen as follows:

1) rank $K_3(R) = r_2 = 1$, the number of pairs of complex embeddings of $F$ (see [W, Theorem 1.5])

2) Since $R$ is euclidian, $SL(R) = E(R)$, the subgroup generated by elementary matrices.

3) According to [A, after Cor. 5.1] the Hurewicz homomorphism $$h_n: K_n(R) = \pi_nBGL(R)^+ \cong \pi_nBE(R)^+ \to H_n(E(R);\mathbb{Z}) = H_n(SL_n(R); \mathbb{Z})$$ is surjective for $n=3$ with torsion kernel.

4) It follows that the ranks of $K_3(R)$ and $H_3(SL(R); \mathbb{Z})$ are equal and by 1) they are equal $1$.

5) $GL(R) \cong SL(R) \times R^\times = SL(R) \times \mathbb{Z} /6 \mathbb{Z}$.

6) From Kunneth formula it follows that $SL(R) \hookrightarrow GL(R)$ induces a monomorphisms $$H_3(SL(R),\mathbb{Z}) \hookrightarrow H_3(GL(R); \mathbb{Z})$$ with torsion cokernel.

7) It follows from 6) and 3) that the Hurewicz homomorphism $$\bar{h}_3: \pi_3(R)/(\text{torsion}) \to H_3GL(R); \mathbb{Z})/(\text{torsion})$$ is an isomorphism and the index that was asked for equals $1$.


a) An explicit description of $K_3(F)$ is given by [W, Theorem 1.7]

b) Note that $K_n(F) = K_n(\mathcal{O})$ for $n$ odd (see [W, Theorem 1.6])


[A] Arlettaz, The exponent of the homotopy groups of Moore spectra and the stable Hurewicz homomorphism

[W] Weibel, Albebraic K-Theory of Rings of Integers in Local and Global Fields

Edit: To complete the story: The Hurewicz map $$h_3: K_3(F) \to H_3(SL(F);\mathbb{Z})$$ also induces an isomorphism on the torsion-quoients and is of index $1$ there.

For, from 1) and b) we know that the torsion free part of $K_3(F)$ has rank $1$ and 3) is also true with $F$ in place of $R$. Since $SL(F) = E(F)$ (this holds for every local ring) the assertin follows from 3).

That $H_3(GL(F);\mathbb{Z})$ is no good choice for approximating $K_3(F)$ was already discussed in the comments above.

share|cite|improve this answer
Thanks very much. Actually, just the fact that the Hurewicz homomorphism, $h_3: K_3(R) \to H_3(E(R);\mathbb{Z}$ is surjective is enough for me. That really helps. – tkf Aug 9 '11 at 17:38

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.