Let $R$ be a a commutative ring with a unit element, then one can associate to $R$ a Boolean ring $B(R)$, as in this text by Bergman, last line of page 594. (I guess this is a very classical thing. Explicitly, $B(R)$ is the set of idempotent elements of $R$ with the operations $e \oplus f = e + f - 2ef$ and multiplication inherited from $R$.) Bergman shows that the functor $R \mapsto B(R)$ has a left adjoint - the one you would expect. I suspect that it does not have a right adjoint, but I don't see how to prove this (if it is true).

All ideas would be welcome!

  • 1
    $\begingroup$ A typical method is to use the fact that if it has a right adjoint, it preserves colimits. (I don't know the situation to know if that will work here) $\endgroup$ – user14972 Oct 8 '12 at 15:01
  • $\begingroup$ Hm... and you're suggesting this is not the case? $\endgroup$ – Evariste Oct 8 '12 at 15:02
  • 1
    $\begingroup$ So that this question can be useful to future surfers, could you, @Hurkyl, or Manny Reyes possibly add a description of the functor $B$? $\endgroup$ – Kevin Carlson Oct 8 '12 at 21:07

I will make use of Hurkyl's comment above. If this functor has a right adjoint, then it must preserve colimits, and in particular, it must send coproducts to coproducts. This is not the case, so the functor does not have a right adjoint.

The coproduct in the category of commutative rings is well-known to be the tensor product $-\otimes_{\mathbb{Z}}-$, which I will henceforth write as $-\otimes-$. This has the funny property that the coproduct of two nonzero rings may be zero. For a famous example, if $R = \mathbb{Z}/(2)$ and $S = \mathbb{Z}/(3)$, then $R \otimes S = 0$. Thus $B(R \otimes S)$ is the trivial Boolean ring, in which $0 = 1$.

On the other hand, because $R$ and $S$ are fields it follows that $B(R)$ and $B(S)$ are both (isomorphic to) $\mathbb{Z}/(2)$. Because $\mathbb{Z}/(2)$ is the initial object in the category of Boolean rings, the coproduct of $B(R)$ and $B(S)$ in this category readily seen to be $\mathbb{Z}/(2)$ once again.

So the functor $B$ applied to the coproduct of these rings $R$ and $S$ is not isomorphic to the coproduct of the Boolean rings $B(R)$ and $B(S)$.

  • $\begingroup$ @Hurkyl, thank you very much for fixing the typo! $\endgroup$ – Manny Reyes Oct 9 '12 at 20:42

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.