Let $f \in \mathbb{C}[x_1,\ldots,x_n]$, $n \geq 1$. Call $f$ 'good' if $\mathbb{C}[x_1,\ldots,x_n]$ is flat over $\mathbb{C}[f]$.
Is it true that every $f \in \mathbb{C}[x_1,\ldots,x_n]$ is good?
- $f \in \mathbb{C}$ is good, since $\mathbb{C}[x_1,\ldots,x_n]$ is free (hence flat) over $\mathbb{C}$.
- $n=1$: $\mathbb{C}[x]$ is a free $\mathbb{C}[x^2]$-module (of rank two), hence flat, so $\mathbb{C}[x]$ is flat over $\mathbb{C}[x^2]$. Similarly, if $f \in \mathbb{C}[x]$ is of degree $d$, then $\mathbb{C}[x]$ is free (hence flat) over $\mathbb{C}[f]$ of rank $d$. Therefore, every $f$ is good.
- $n=2$: If $f \in \mathbb{C}[x]$ then $\mathbb{C}[x,y]$ is free (of infinite rank) over $\mathbb{C}[f]$ and if $f \in \mathbb{C}[y]$ then $\mathbb{C}[x,y]$ is free (of infinite rank) over $\mathbb{C}[f]$. But what about a general $f \in \mathbb{C}[x,y]$? It seems to me that the extension is still free: If $\deg_x(f)=d_1 < \infty$ and $\deg_y(f)=d_2 < \infty$, then $\mathbb{C}[x,y]$ is free of rank $d_1d_2$. If $\deg_x(f) = -\infty$ or $\deg_y(f) = -\infty$, then $\mathbb{C}[x,y]$ is free of infinite rank.
- $n \geq 3$, similarly to the $n=2$ case, every $f$ is good.
Summarizing, it seems that $\mathbb{C}[x_1,\ldots,x_n]$ is free (hence flat) over $\mathbb{C}[f]$, for every $f \in \mathbb{C}[x_1,\ldots,x_n]$.
Please, am I right or wrong? Thank you very much.
Here is an algebraic-geometric way to prove this, though it might be considered cracking a nut with a sledgehammer. For non-constant $f$, the inclusion $\mathbb C[f] \subset \mathbb C[x_1, \dotsc, x_n]$ induces a morphism of schemes $$f: \mathbb A^n \to \mathbb A^1, \quad (p_1, \dotsc, p_n) \mapsto f(p_1, \dotsc, p_n).$$ Hartshorne's Algebraic Geometry, Chapter III.9 contains the following
As $\mathbb A^n$ is reduced with a single irreducible component, we see that the morphism $f$ is flat because $f$ is non-constant.