Let $R$ be a spectrum. Assume that $R$ is bounded-below. Then we can “even-ify” $R$: cone off some generating set of the first nonvanishing odd-dimensional homotopy group of $R$. Do this repeatedly. In the end you have a spectrum $R^{ev}$ with even-dimensional homotopy. Moreover, you only added even cells, so if $R$ had even cells to begin with, then $R^{ev}$ has even cells.

If $R$ is a ring spectrum, then so is $R^{ev}$, by obstruction theory.

Question 1: If $R$ is a connective ring spectrum with even cells, then is the homotopy of $R^{ev}$ zero-divisor-free?

Question 2: If $R$ is a ring spectrum with even homotopy and even cells, then is $\pi_\ast R$ zero-divisor-free?


  • I’m actually not quite sure how to ensure that $R_\ast$ is a commutative ring, so perhaps the title question — asking if $R_\ast$ is in fact polynomial — doesn’t quite make sense.

  • For example, when $R$ is the sphere, I think we get $R^{ev} = MU$. At any rate, if we work $p$-locally and take $R$ to be the $p$-local sphere, then I’m certain I’ve been told that $R^{ev} = BP$. So an abstract proof that even-ification produces polynomial homotopy groups might provide some alternate way to think about the Milnor/Quillen computation of the homotopy groups of $MU$.

  • If we do this unstably, then the possible output spaces were completely classified by Steve Wilson in his PhD thesis. I think the theorem says they are all products of spaces of the form $\Omega^{\infty-n}BP$. At any rate, the list is very finite.

  • Note that if we don’t require $R$ to have even cells, then there are Eilenberg-MacLane counterexamples (Just take $HR$ where $R$ is an even-graded ring with zero-divisors). But I’m pretty sure that an Eilenberg-MacLane spectrum never has even cells.

Question 3: Is there a classification of (ring) spectra with even homotopy and even cells analogous to Wilson’s classification of spaces with even homotopy and even cells?


1 Answer 1


Let $G$ be any discrete group, and let $MU[G] = MU \otimes \Sigma^\infty_+ G$ be the associated group algebra over $MU$. Additively, $MU[G] \simeq \bigoplus_{g \in G} MU$, and so it has both even homotopy and is built from even cells - it is (one choice of) its own evenification. However, its homotopy typically has zero-divisors (if $g^n = 1$, then $(1-g)(1+g+\dots+g^{n-1}) = 0$).

Another example is given by the trivial square-zero extension $MU \oplus \Sigma^{2n} MU$, whose coefficient ring is $MU_*[x] / x^2$ for a generator $x$ in degree 2n.

I don't have a good answer for question 3. A classification of such objects seems like it would be very difficult because of a wide variety of examples.

(The identification of the evenification of $BP$ is due to Priddy, "A cellular construction of BP and other irreducible spectra". In the integral case there is not usually a canonical minimal construction of $R^{ev}$ like there is in the $p$-local case.)


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.