Hyperplane $\Phi_X$ in Chow construction (Algebraic Geometry by Joe Harris)

84 Views Asked by At

I'm trying to figure out what Harris wanted to say in following construction, called Chow construction (Algebraic Geometry by Joe Harris, p. 269):

The first construction of the parameter space for varieties of a given degree and dimension is called the Chow construction. The basic idea behind it is simply that the problem in parametrizing varieties $X \subset \mathbb{P}^n$ in general is that $X$ is not generally given by a single polynomial, whose coefficients we can vary freely as in the example of hypersurfaces. The point of the construction is thus to associate to any such variety $X$ a hypersurface $\Phi_X$, albeit not one in a projective space.

There are two essentially equivalent ways of doing this. Suppose first of all that $X$ has pure dimension $k$, and consider the incidence correspondence consisting of points $p \in X$ together with $(k + 1)$-tuples of hyperplanes containing $p$; that is,

$$ \Gamma = \{(p, H_1,..., H_{k+1}): \ p \in H_i \forall i \} $$ $$ \subset X \times (\mathbb{P}^{n})^* \times ... \times (\mathbb{P}^{n})^* $$

($(\mathbb{P}^{n})^*$ is the dual projective space) for each point $p \in X$, the set of hyperplanes containing $p$ is a hyperplane $\mathbb{P}^{n-1} \subset (\mathbb{P}^{n})^* $, so that the set of $(k + 1)$-tuples of hyperplanes is irreducible of dimension $(k + 1) \cdot (n - 1)$. We deduce that $\Gamma$ is of pure dimension $k + (k + 1) \cdot (n - 1) = (k + 1)· n - 1$, with one irreducible component corresponding to each irreducible component of $X$.

For a general choice of point $p \in X$ and $H_1,..., H_{k+1}$ containing $p$ the intersection of the $H_i$ with $X$ will consist only of the point $p$, so that the canonical projection map $\pi: \Gamma \to (\mathbb{P}^{n})^* \times ... \times (\mathbb{P}^{n})^* $ will be birational. It follows that the image of $\Gamma$ under this projection is a hypersurface in $(\mathbb{P}^{n})^* \times ... \times (\mathbb{P}^{n})^* $ (this is where we need X to have pure dimension); we will call this hypersurface $\Phi_X$.

Question: The last sentences I not understand. If a map between $f: X \to Y$ varieties/ schemes of pure dimension is birational then the image (and also it's closure in $Y$) cannot be contained in a subvariety $W \subset Y$ of smaller dimension than $\dim Y (= \dim \ X$ as $f$ birational). But a hypersurface of $Y$ has properly smaller dimension than $Y$ so the map cannot be birational.

Therefore I not understand why the projection map $\pi: \Gamma \to (\mathbb{P}^{n})^* \times ... \times (\mathbb{P}^{n})^* $ is birational and a hypersurface of $(\mathbb{P}^{n})^* \times ... \times (\mathbb{P}^{n})^* $ contains the image of $\Gamma$.