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

In group cohomology, for $H$ a finite-index subgroup of $G$ and $M$ a $G$-module, there is a transfer (or corestriction) map $Cor : H^* (H;M) \to H^*(G;M)$.

In homotopy theory, there is a transfer map for finite covering spaces $\bar{X} \to X$, and it exists for all coefficient systems on $X$. It is given on homology (say) by sending a small simplex in $X$ to its finitely-many lifts to $\bar{X}$. The group transfer is obtained from the topological one by the finite covering space $BH \to BG$.

There is a fancier transfer, due to Becker and Gottlieb, for any map $E \to B$ whose homotopy fibre is stably equivalent to a finite complex. Can this be extended to give a transfer map for coefficient systems on $B$?

share|cite|improve this question
Hi Oscar. If your map $E\to B$ is smooth, there is a Gysin map with local coefficients, described here:… – Mark Grant Nov 11 '10 at 10:45
up vote 6 down vote accepted

I believe the answer is yes. The idea should be to generalize the construction of the group-theoretic transfer. I'll describe how I think this ought to go. I'm pretty sure what I'm describing is a known construction, but I haven't been able to dig up a reference which describes it.

Let $f:X\to Y$ be a map of spaces (CW-complexes), and assume (for simplicity) that both are path connected. Then there is an inclusion of topological groups $\phi: G\to H$, such that $B\phi: BG\to BH$ is equivalent to $f:X\to Y$.

A local system on $X$, in the most general possible context, is a spectrum $M$ equipped with a $G$-action. (I'm not doing anything fancy here; in particular, I'm not doing equivariant stable homotopy theory. Just spectra with a $G$-action; a $G$ map $f:M\to N$ of such things is a weak equivalence if it is a weak equivalence of the underlying spectra.)

Let $S_G$ be the category of $G$-spectra. It'll be important to note that this is a closed monoidal category: the smash product $X\wedge Y$ of two objects in $S_G$ is defined in $S_G$ (smash the spectra, and use the diagonal $G$-action), as well as a function object $\mathrm{Hom}(X,Y)$ (take the spectrum function object, with diagonal $G$-action).

The Becker Gottlieb transfer $f_\!$ of $f$ associated to an $M\in S_H$ should be a map $M_{hH}\to M_{hG}$ (where "$hG$" is "homotopy orbits"). So, if $M=S^0$ with trivial $H$-action, you get a map $S^0_{hH}\to S^0_{hG}$, which is a map $\Sigma^\infty Y_+\to \Sigma^\infty X_+$. You should interpret the map $\pi_* f_\!$ as being a map $$ H_*(Y,M) \to H_*(X,f^*M), $$ where these are "homology with coefficients in the local systems $M$ and $f^*M$". There should be a similar way to get a cohomolgy transfer.

Let $F=\Sigma^\infty(H/G)_+$, as a spectrum with $H$-action. Note that the space $H/G$ is just the fiber of $f:X\to Y$.

Here is a sequence of maps in $S_H$: $$ M\to \mathrm{Hom}(F,F\wedge M) \leftarrow \mathrm{Hom}(F,M)\wedge F \to \mathrm{Hom}(F,M)\wedge F\wedge F \to M\wedge F \approx M\wedge_G H. $$ The first is smashing with the identity map of $F$; the second comes from smashing a map $F\to M$ with a map $S^0\to F$; the third comes from the diagonal map $H/G\to H/G\times H/G$; the fourth is "evaluation".

If $F$ has the homotopy type of a finite CW-spectra (i.e., if $H/G$ is finitely dominated), then the backwards map in this sequence is a weak equivalence. In this case, we get a map $M\to M\wedge_G H$ of $H$-spectra, and taking homotopy orbits gives $M_{hH}\to M_{hG}$.

If $Y$ is contractible, this amounts to a map $$ S^0 \to \mathrm{Hom}(F,F) \leftarrow \mathrm{Hom}(F, S^0)\wedge F \to \mathrm{Hom}(F, S^0)\wedge F\wedge F \to S^0\wedge F, $$ which is to say, a map $S^0\to \Sigma^\infty X_+$, and in cohomology this will send $1\in H^0(X)$ to $\chi(X)\in H^0(\mathrm{point})$. So apparently we recover the Becker-Gottlieb transfer.

If the spaces aren't connected, you can still do all this, but you need to allow $H$ and $G$ to be groupoids.

(I learned this way of thinking from some paper of John Klein about the "dualizing spectrum" of a toplogical group; I can't find one where he addresses the BG transfer this way.)

share|cite|improve this answer

I think what Charles has written up above is closely related to what Becker and Gottlieb did in

Becker, J. C.; Gottlieb, D. H. Transfer maps for fibrations and duality. Compositio Math. 33 (1976), no. 2, 107–133.

They work fiberwise, but that is equivalent to working equivariantly in the sense that a $G$-space $X$ gives rise to a fibration $X \times_G EG \to BG$ (Borel construction; this induces a Quillen equivalence).

share|cite|improve this answer

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.