For the purposes of this question, the Friedlander-Milnor (FM) conjecture asserts an equality of the group homology for algebraic groups, and their discretizations in the following sense:

Conjecture FM: For a reductive algebraic group $G$ over a separably closed field $k$, the map $BG^\delta \to BG$ induces an isomorphism on $H^*_{\mathrm{et}}(-,\mathbf{Z}/\ell)$ for any $\ell$ invertible on $k$; here $G^\delta = G(k)$ with the discrete topology.

What is the status of FM today? Here is what google tells me:

1) Friedlander and Mislin proved the conjecture for $k = \overline{\mathbf{F}_p}$ in their 1984 paper "Cohomology of classifying spaces of complex Lie groups and related discrete groups."

2) There is a 2011 paper by Morel showing special cases of this result, and establishing a connection with $\mathbf{A}^1$-homotopy theory by proving the so-called "weak Friedlander-Milnor" conjecture.

In particular: is there a single example of a field $k$ of characteristic $0$ where we know FM?


1 Answer 1


Here is what I know about the status of the question (as of today):

At the moment, there is no field of characteristic 0 for which the Friedlander-Milnor conjecture is known for a semisimple algebraic group. From Matsumoto's theorem, it follows for $H_2$ of semisimple groups, from the work of Suslin it follows in the stable range and for the limit groups $GL_\infty$, $Sp_\infty$ and $O_\infty$, and from the work of Sah it follows for $H_3(SL_2)$. And of course, Friedlander and Mislin proved it for the algebraic closure of a finite field. Those results can be found in the book "Homology of linear groups" by K. Knudson.

As far as I know, and modulo checking some details, what I think Morel establishes in the paper (via a construction of transfers on $\mathbb{A}^1$-classifying spaces of split semisimple linear algebraic groups) is the equivalence of the following statements about change-of-topology maps (whenever they make sense):

  1. weak homotopy invariance, i.e., the canonical map $G(k)\to G(k[\Delta^\bullet])$ induces an isomorphism in mod $\ell$ homology, where $k[\Delta^\bullet]$ is the standard polynomial cosimplicial object.

  2. Milnor's version of the Friedlander-Milnor conjecture, i.e., the canonical change-of-topology map $G^\delta\to G$ induces an isomorphism in mod $\ell$ homology, for $G$ a Lie group.

  3. Friedlander's generalized isomorphism conjecture, i.e., the canonical map $BG(k)\to BG^{\operatorname{et}}_k$ from discrete to the étale classifying space induces an isomorphism in mod $\ell$ homology, for $G$ a linear algebraic group over an algebraically closed field $k$.

The original proof strategy then was to prove homotopy invariance, which however has turned out to be a lot more subtle than expected: Theorem 1.2 in the 2009 paper is wrong, and Theorem 7 in the 2011 paper of Morel is not known at the moment (partly my fault). In fact, a strong form of homotopy invariance does not seem to hold for arbitrary semisimple groups. I wrote a paper on counterexamples to homotopy invariance, which is published as:

M. Wendt. On homotopy invariance for homology of rank two groups. J. Pure Appl. Algebra. 216(10), 2012, pp. 2291--2301.

Even worse, weaker forms of homotopy invariance also seem to generally fail. There is another paper

K. Hutchinson and M. Wendt. On third homology of $SL_2$ and weak homotopy invariance. ArXiv:1307.3069.

In this paper, there are many examples of fields where the natural map $H_3(SL_2(k),\mathbb{Z})\to H_3(BSL_2(k[\Delta^\bullet]),\mathbb{Z})$ fails to be an isomorphism. The counterexamples known at the moment all depend on non-triviality of square classes, so they do not say anything about quadratically closed fields.

The current status is that there is a new announcement on Morel's webpage claiming a proof of weak homotopy invariance for $SL_2$ over quadratically closed fields. This together with the rigidity result in his 2011 paper would then imply the Friedlander-Milnor conjecture for $SL_2$ - both Milnor's form of the conjecture for $SL_2(\mathbb{C})$ and Friedlander's generalized isomorphism conjecture for $SL_2$ over any algebraically closed field. Milnor's form of the conjecture for $SL_2(\mathbb{R})$ would not follow from the announced result as $\mathbb{R}$ is not quadratically closed.

  • 1
    $\begingroup$ I notice that there's another paper by Amrani which seeming claims the statement for $k=\mathbb{C}$? So is this work accepted by the community or is it faulty? $\endgroup$
    – XYC
    Jul 23, 2022 at 17:13

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.