I have a cocomplete abelian category $\mathcal C$ and two objects $X$, $Y$ in $\mathcal C$. Further, $\mathcal C$ has a projective generator $P$. I have an isomorphism

$$ \mathcal C(P,X) \cong \mathcal C(P,Y)$$

It feels like this should be enough to show that $X\cong Y$ in $\mathcal C$. Because any object $Z$ in $\mathcal C$ can be written as a cokernel $P^J\longrightarrow P^I\longrightarrow Z\longrightarrow 0$ and so we have $\mathcal C(Z,X)\cong \mathcal C(Z,Y)$. It feels like the isomorphisms $\mathcal C(Z,X)\cong \mathcal C(Z,Y)$ are natural in $Z$ and do not depend on the choice of projective resolutions.

But I cannot be sure. Is this true? Any reference or pointers would be appreciated.

  • $\begingroup$ Your claim is false. You can find a counterexample where $\mathcal{C}$ is the category of real vector spaces. $\endgroup$ – Zhen Lin Jun 15 at 4:45
  • 5
    $\begingroup$ Isomorphism of $\mathcal{C} (P,X)$ with $\mathcal{C} (P,Y)$ as what? If you consider them as $\text{End}(P)^{\text{op}}$-modules, then yes, this implies the isomorphism of objects. $\endgroup$ – Sasha Jun 15 at 8:10

It is very easy to specify an answer if $P$ is a compact projective generator (you didn't write whether it is compact). Then, we have an equivalence of categories between $\mathcal{C}$ and the category of right modules over $\text{End} (P)$, given by sending $X$ to $\mathcal{C} (P , X)$. So, of course, if objects become isomorphic under an equivalence of categories, they were isomorphic. For that answer, one also needs to understand your isomorphism as compatible with the endomorphisms of $P$, i.e. a $\text{End}(P)$-module isomorphism.


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.