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 me first recall some basic well-known definitions: Let $R$ be a ring (as always commutative). A (cocomplete) abelian tensor category is defined to be a symmetric monoidal category, whose underlying category is also a $R$-linear (cocomplete) abelian category, such that the tensor product is right exact (cocontinuous) and $R$-linear in each variable. An object $X$ of such a category $C$ is called flat if $X \otimes -$ is an exact functor. Also recall that an object $X$ of a symmetric monoidal category is called dualizable with dual $X^\*$ if it comes equipped with morphisms $1_C \to X \otimes X^\*$ and $X^\* \otimes X \to 1_C$ satisfying the two triangular identities.

Now these notions are connected in the case of $C = \text{Mod}(R)$: The dualizable objects are precisely the finitely generated projective $R$-modules (equivalently, locally free of finite rank). Thus, every dualizable object is flat, and Lazard's Theorem implies that every flat module is a directed colimit of locally free modules. Now I wonder if these statements remain true in the general case. I've already this concerning Lazard's theorem here for the category of quasi-coherent modules on a nice scheme (without answer). Thus my question is:

Question: Is every dualizable object in a (cocomplete) abelian tensor category flat?

Remark that we cannot insert "projective" as an intermediate step. It is easy to see that $1$ is projective iff every dualizable object is projective. But $1$ is not projective in $\text{Qcoh}(S)$ forr some non-affine concentrated scheme $S$, whereas the question in this case is answered with "yes" since every locally free quasi-coherent sheaf is locally flat and thus (globally) flat.

share|cite|improve this question
Thank you Nacho and Todd for your answers. I would like to accept them both ;) – Martin Brandenburg Jun 24 '11 at 12:45
up vote 15 down vote accepted

A dual pair as you describe above induces an adjunction $(X^*\otimes-)\dashv(X\otimes-)$, and since right adjoints preserve limits, $X$ is flat.

share|cite|improve this answer
Yes of course! I was, again, blind. – Martin Brandenburg Jun 24 '11 at 12:44
Indeed, this proves something much more than flatness for $X$: namely, $X\otimes$ preserves all limits, not just the finite ones. (I've usually seen the word "exact" meaning only preservation of finite limits and colimits.) – Theo Johnson-Freyd Jun 24 '11 at 13:49

In a symmetric monoidal category, $X$ and $X^\ast$ are mutually dual. (We can compose with the symmetry to obtain a unit $1_C \to X^\ast \otimes X$ and counit $X \otimes X^\ast \to 1_C$ which realize $X$ as the dual of $X^\ast$.) This means we have both $X \otimes - \dashv X^\ast \otimes -$ and $X^\ast \otimes - \dashv X \otimes -$. Since $X \otimes -$ is both a left and right adjoint, it is exact.

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.