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 $R$ be a reduced algebra of finite type over a field $k$ of characteristic 0. Let $S$ be a reduced finite $R$-algebra. Is $S \otimes_R S$ reduced?

(In positive characteristic one can get non-reduced tensor products of reduced algebras even over a field.)

I have failed to find a counterexample so I thought that the statement might be true after all. The question is motivated by the discussion in the comments to this question.

If $S=R[x_1, \ldots, x_n]/I$, what one has to show is that the ideal $I \otimes 1 + 1 \otimes I$ in $S \otimes_R S=R[x_1, \ldots, x_n, y_1, \ldots, y_n]$ is radical. However, I have no good ideas as to how to approach this.

share|cite|improve this question
up vote 13 down vote accepted

No. Let $R = k[x,y]/(y^2-x^3)$. Let $S = k[t]$, with the map $R \to S$ given by $(x,y) \mapsto (t^2, t^3)$. So $S \otimes_R S = k[t,u]/(t^2-u^2, t^3-u^3)$. The element $t-u$ is not zero in the tensor product, because all the relations are in degree $>1$. But $(t-u)^3 = 3 (t-u)(t^2-u^2) - 2 (t^3-u^3)$ is in the ideal.

share|cite|improve this answer
Thank you! I wonder if it is a coincidence that this counterexample is not flat. – Dima Sustretov Feb 13 '12 at 20:21
Dear David, is it possible to turn this counterexample into a counterexample where S is of the form $R[x_1,\ldots,x_n]/I$ where I is generated by elements linear in $x_i$-s? E.g. applying the functor $Sym$ to $S$ (seen as a finite module over $R$) we get $R[u,v]/(ux−vy)$. However, I can't easily see if $R[u,v,t,s]/(ux−vy,tx−sy)$ is reduced or not. – Dima Sustretov Feb 15 '12 at 14:03
That's reduced. The ideal has primary decomposition $(ux-vy, tx-sy,us-tv) \cap (x,y)$, and each of those are prime. I don't see any fundamental reason something like this couldn't work, though. – David Speyer Feb 15 '12 at 14:30
I think I see why this can't work. For any point $\mathrm{Spec}(K) \in \mathrm{Spec}(R)$, if the equations are of the form you describe, then the fiber of $\mathrm{Spec}(S)$ over $\mathrm{Spec}(K)$ is a linear space. So the fiber of $\mathrm{Spec}(S \otimes_R S)$ over $\mathrm{Spec}(K)$ is the square of that linear space, and is hence reduced. There should be a theorem that, if $Y$ is reduced, and every fiber of $X \to Y$ is reduced, then $X$ is reduced. I don't know a citation for that, though. – David Speyer Feb 15 '12 at 14:43
This is nice. Concerning the theorem that you mention, is $Y$ over a field or is it a general scheme? – Dima Sustretov Feb 15 '12 at 15:23

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.