Why is a smooth scheme over a field reduced

algebraic-geometrycommutative-algebra

I think I have some fundamental misunderstanding of smoothness of scheme now. It is mentioned in many places that "a scheme smooth over a field is reduced" (which, of course, also enjoys other nice properties.). But now consider the situation that $X=\text{Spec }k[t]/(t^n) $ and $Y=\text{Spec }k$, where $k$ is a field. (Let's say it has characteristic 0.) Let $f:X\to Y$ be the natural map. Then, following the definition in Hartshorne's book,

  1. $f$ is flat.
  2. $\dim X=\dim Y+0$.
  3. $\Omega_{X/Y}=0$

Thus, $f$ is smooth of relative dimension 0. But I think it should not be smooth. What is my misunderstanding?

Best Answer

$\Omega_{X/Y}\neq 0$. In particular, the module of differentials $\Omega_{(k[t]/(t^n)/k)}$, which is the global sections of $\Omega_{X/Y}$, is a $k[t]/(t^n)$ module with generator $dt$ and relations generated by $nt^{n-1}dt=0$. So $dt$ is nonzero when $n>1$.

Related Question