4

I am considering the coordinate ring $k[X \times\mathbb{A}^n]$, where $X$ is an algebraic variety in $\mathbb{A}^n$. I want an isomorphism between this and the polynomial ring $k[X][y_1,\ldots, y_n]$. I have proven that $X \times\mathbb{A}^n$ is indeed a variety, but I am unsure of where to go from here. The only information I have found seems to do with tensor products, which I am unfamiliar with.

Is it necessarily true that $k[X \times\mathbb{A}^n] = k[x_1,\ldots, x_{2n}]/ Id(X\times \mathbb{A}^n)$ ? I am trying to manipulate this....

EDIT: I think I have it, but I'm making the assumption that $Id(X\times\mathbb{A}^n) = Id(X) \times Id(\mathbb{A}^n)$. Is this valid?

Joe
  • 185
  • 10
  • 2
    In general, if $X$ and $Y$ are affine varieties, then the coordinate ring of $X \times Y$ is the tensor product $k[X] \otimes_k k[Y]$. If you aren't familiar with tensor products this is an excellent excuse to learn what they're all about. – Qiaochu Yuan Feb 05 '13 at 07:36

1 Answers1

1

Let us call $x_1,\dots,x_n$ the coordinates on $X$ when viewed inside $\mathbb A^n$, and $y_1,\dots,y_n$ the coordinates on the other copy of $\mathbb A^n$ in the product $X\times\mathbb A^n$. Then, by definition, $$k[X\times\mathbb A^n]=k[x_1,\dots,x_n,y_1,\dots,y_n]/I(X\times \mathbb A^n),$$ where $I(X\times \mathbb A^n)\overset{(\ast)}{=}\sqrt{I(X)+I(\mathbb A^n)}=\sqrt{I(X)}=I(X)$. (I assume in the last equality that your variety is reduced. See e.g. here for $(\ast)$)

It is true that, using tensor product, you get the result. Indeed, $$ k[X\times\mathbb A^n]\overset{(\ast\ast)}{=}k[X]\otimes_kk[\mathbb A^n]=k[X]\otimes_kk[y_1,\dots,y_n]\overset{(\sharp)}{\cong}k[X][y_1,\dots,y_n]. $$

For equality $(\ast\ast)$ you might look at the same page as before. For $(\sharp)$, it is a general fact that tensoring a $k$-algebra $B$ with a polynomial algebra $A:=k[y_1,\dots,y_n]$ gives you $B[y_1,\dots,y_n]$.

Here is a proof. One verifies that $B[y_1,\dots,y_n]$ satisfies the universal property of the tensor product $A\otimes_kB$. Indeed, one has two morphisms $A\to B[y_1,\dots,y_n]\leftarrow B$, and any other couple of morphisms $A\to C\leftarrow B$ factors uniquely through some $p:B[y_1,\dots,y_n]\to C$. Try to define such $p$.

However, it would be easy to show (without tensor products) that $k[x_1,\dots,x_n,y_1,\dots,y_n]/I(X)=k[x_1,\dots,x_n][y_1,\dots,y_n]/I(X)\cong k[X][y_1,\dots,y_n]$, because $y_1,\dots,y_n$ do not "intervene". You may try to define this isomorphism explicitly.

Brenin
  • 14,072