Let $X$ be a complex projective and smooth toric variety of complex dimension $n$. It is acted by the real torus $T=(S^1)^n$.

Is it true that the $T$-equivariant cohomology $H^*_T(X,\mathbb{Z})$ of $X$ is isomorphic as a graded algebra to $$H^*(X,\mathbb{Z})\otimes H^*_T(pt,\mathbb{Z})\simeq H^*(X,\mathbb{Z})\otimes \mathbb{Z}[x_1,\dots,x_n],$$ where $x_i$ have degree 2? (to prove such an isomoprhism with coefficients in a field would be fine too.)

Remark. I think I have a proof of such an isomorphism (with $\mathbb{Q}$-coefficients) in the category of graded $H_T^*(pt, \mathbb{Q})$-modules rather than graded algebras (which seems to be well known).

  • 5
    $\begingroup$ Isn't this false for $\Bbb P^1$? There we have $H^*_{S^1}(\Bbb P^1) = H^*(\Bbb{CP^\infty} \vee \Bbb{CP}^\infty)$ (as that wedge is the Borel construction on $\Bbb P^1$) which has cohomology $\Bbb Z[x,y]/(xy)$. In particular, it has no nilpotent elements, unlike your tensor product. $\endgroup$ – Mike Miller Dec 25 '19 at 16:19
  • 2
    $\begingroup$ As graded $H^\ast_T(pt,\mathbb Q)$-modules it is just the fact, that the action of $T$ on $X$ is Hamiltonian with respect to the invariant Kähler structure. This implies that the action is equivariant formal, which means your isomorphism is an isomorphism of $H^\ast_T(pt,\mathbb Q)$-modules. $\endgroup$ – Panagiotis Konstantis Dec 25 '19 at 17:14
  • $\begingroup$ @MikeMiller: Why not make your comment into an answer? I'd be interested to know how you see the homotopy equivalence you mentioned. $\endgroup$ – Mark Grant Jan 6 '20 at 12:01

As per Mark Grant's suggestion, here are some additional details for my comment.

We may write $\Bbb P^1$ with its circle action as the adjunction space $$* \cup_{S^1 \times 0} S^1 \times [0,1] \cup_{S^1 \times 1} *,$$

with the circle acting in the obvious way on $S^1 \times [0,1]$. Passing to the Borel construction, we find that $$\Bbb P^1 \times_{S^1} ES^1 = BS^1 \cup_{ES^1 \times 0} ES^1 \times [0,1] \cup_{ES^1 \times 1} BS^1.$$

The picture is that we are connecting two copies of $BS^1$ by a contractible bit.

To make this precise, collapse the closed contractible subspace $ES^1 \times 1/2$ to a point. As this is contractible, the projection map is a homotopy equivalence; thus $\Bbb P_{hS^1}$ is homotopy equivalent to the wedge of two mapping cones $$\text{Cone}(ES^1 \to BS^1) \vee \text{Cone}(ES^1 \to BS^1).$$ Because $ES^1$ is contractible, each of these cones are homotopy equivalent to $BS^1$ itself.

Thus we see that $$\Bbb{CP}^\infty \vee \Bbb{CP}^\infty \simeq \Bbb P^1_{hS^1};$$ further the projection map $\Bbb P^1_{hS^1} \to \Bbb{CP}^\infty$ sends the two wedge summands identically onto $\Bbb{CP}^\infty$.

Thus $$H^*_{S^1}(\Bbb P^1;\Bbb Z) \cong \Bbb Z[x,y]/(xy),$$ with action of $u \in H^2_{S^1}(pt)$ given by $u \cdot x = x^2, u \cdot y = y^2.$ This is isomorphic as a graded module to $$H^*(\Bbb P^1;\Bbb Z) \otimes H^*_{S^1}(pt;\Bbb Z),$$ but not as an algebra: the equivariant cohomology has no nilpotent elements, whereas the tensor-product algebra does, given by the generator of $H^2(\Bbb P^1) \otimes H^0_{S^1}(pt)$.

  • $\begingroup$ A previous version of this answer just wrote down a map $$\Bbb{CP}^\infty \vee [0,1] \vee \Bbb{CP}^\infty \hookrightarrow \Bbb P^1_{hS^1}$$ which Mayer-Vietoris and van Kampen show is an iso on $\pi_1$ and homology, so is a homotopy equivalence by Hurewicz and Whitehead. This seemed cleaner, though. $\endgroup$ – Mike Miller Jan 6 '20 at 16:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.