In Stack project Definition $29.32.13$ of smoothness there is a counter-example of a morphism $f:\operatorname{Spec} \Bbb F_p[t] \to \operatorname{Spec} \Bbb F_p[t^p]$ that is:
- flat
- locally of finite presentation
- the module of differentials is locally free
But f is not smooth.
Trying to work out the details:
- flat
Denote $X:=\operatorname{Spec} \Bbb F_p[t^p], Y:=\operatorname{Spec} \Bbb F_p[t]$. So $f:Y \to X$ induces $g: A\to B$ with $A:=\Bbb F_p[t^p], B:=\Bbb F_p[t]$
$\Bbb F_p[t^p]$ is PID as $\Bbb F_p$ is a field, and $B$ is a torsion-free A-module so $B$ is flat A-module, hence $f$ is flat. Is this correct?
- locally of finite presentation (Solved)
I tried to find a surjection $\Bbb A[t_1,...,t_n] \to B$ with finitely generated kernel but couldn't. I have hard time visualizing what kind of morphisms $A\to B$ can exist.
- the module of differentials is locally free (solved)
$\Omega_{B|A}$ is generated by elements $(d b)_{b\in B}$, with $d:b\to \Omega_{B|A}, b \to 1\otimes b - b \otimes 1$. But B is commutative so $\Omega_{B|A} = 0$. Does it follow that $\Omega_{Y|X}$ is locally free?.
- Finally for smoothness, I want to show that the fibers are not equidimensional. Let $x\in X$.
$f^{-1}(x)=B\otimes_A\kappa(x)$. I couldn't compute residue field $\kappa(x)$ as I couldn't describe exactly elements of X.
Thank you for any help.