Let $\phi: A=\mathbb{C}[x,y]/(y^2-x^3) \to \mathbb{C}[t]=B $ be the ring morphism defined by $x\mapsto t^2, y \mapsto t^3.$ Let $f:Y=\text{Spec} B \to X=\text{Spec} A$ be the associated morphism of affine schemes.

Since $\phi$ is injective, then $f^{\sharp} : \mathcal{O}_X \to f_{\star}\mathcal{O}_Y$ is injective too.

1.What is the cokernel of $f^{\sharp}?$

I know that it will be isomorphic to the quotient sheaf $f_{\star}\mathcal{O}_Y/\mathcal{O}_X$ but I'm unable to find it explicitly.

Here I computed the cotangent sheaf $\Omega_X$ of $X.$

2.How can I find the kernel and the cokernel of $f^{\star} \Omega_X \to \Omega_Y?$

What I've tried out:

The sheaf morphism corresponds to a $B$-module morphism $\Omega_A \otimes_A B \to \Omega_B$ or $Adx \oplus Ady/(2ydy-3x^2dx) \otimes B \to Bdt$ which is in fact, the map $(t)dt \oplus (t^2)dt \to Bdt$ where $(tg(t)dt,t^2h(t)dt)\mapsto (tg(t)+t^2h(t))dt$ for $g,h \in B.$ The kernel is isomorphic to $B$ since $g(t)=-th(t)$ and the cokernel is $B/(t).$

Is this correct?


Hopefully you know that coherent sheaves over $\mathrm{Spec} A$ are equivalent to finitely generated $A$-modules. The structure sheaf $\mathcal O_X$ corresponds to $A$ as a module over itself and $f_\ast\mathcal O_Y$ corresponds to $\mathbb C[t]$ with the $A$-module structure given by restricting through $\phi$.

So $f^\#$ corresponds to $\phi$ as a map of $A$-modules, the image of $\phi$ consists of all polynomials with no degree $1$ term so the cokernel is $(t)/(t^2)$. The $A$-module action on this is that $x$ and $y$ act as $t^2$ and $t^3$ respectively, hence $x$ and $y$ act as $0$. So the cokernel is a trivial $1$-dimensional $A$-module.

Edit: I forgot about $1$, the cokernel should be $1$-dimensional, not $2$.

  • $\begingroup$ By $(t)$ you mean the $A$-module generated by $t$ right? $\endgroup$ – Ehsan M. Kermani Feb 4 '13 at 3:54
  • $\begingroup$ By $(t)$ I meant the principal ideal generated by $t$ in $\mathbb C[t]$. $\endgroup$ – Jim Feb 4 '13 at 4:38
  • $\begingroup$ So why the cokernel is not $\mathbb{C}[t]/(t^2)?$ $\endgroup$ – Ehsan M. Kermani Feb 4 '13 at 5:42
  • $\begingroup$ Because the unit $1 \in A$ maps to the unit $1 \in \mathbb C[t]$. So when taking the cokernel you have to factor that out. $\endgroup$ – Jim Feb 4 '13 at 5:45
  • 1
    $\begingroup$ Yes, but that $A$-module homomorphism is exactly the map $\phi$ and because $\phi$ also happens to be a ring homomorphism we know that it sends $1 \mapsto 1$. $\endgroup$ – Jim Feb 4 '13 at 6:03

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.