Differential of defining Equation System (from David Mumfords Complex Algebraic Geometry)

65 Views Asked by At

In David Mumford's book Algebraic Geometry I, Complex Projective Varieties the proof of Bezout's Theorem (5.16 page 80) contains terminology I not understand. Mumford speaks at pages 83 & 84 about 'differentials of equation system (that defines a variety)'. I know what differentials of regular functions are, but I don't know what kind of thing is a 'differential of an equation system ' in classical complex algebraic geometry. We assume:

Let $X^r,Y^s \subset \mathbb{P}^n$ complex algebraic varieties of dimensions $r, s$ with $r+s=n$. Let $L \subset \mathbb{P}^n$ a linear subspace with $L \cap X = \emptyset$. After a reasonable choice of cordinates we can assume $L= V(X_0,...,X_r)$ where $X_i \in \mathbb{C}[X_0,...,X_n]$.

Then the rational map

$$ \sigma:(\mathbb{P}^n - L) \times \mathbb{C} \to \mathbb{P}^n - L$$

def by

$$ \sigma(x_0,...,x_n, t) = (x_0,...,x_r,t x_{r+1},..., t x_n)$$

restricts to regular map $\text{res } \sigma: X \times \mathbb{C} \to \mathbb{P}^n$. Define the algebraic set

$$ \mathcal{X} = \{(x,y,t) \ \vert \ \sigma_t x=y \} \subset X \times Y \times \mathbb{C} $$

and consider the restriction of the $3$rd projection $\pi: \mathcal{X} \to \mathbb{C} $. Locally, $\mathcal{X}$ is defined by $n$ equations:

$$ \tag{EQ} X_i (\sigma_t x) = X_i (y), \ 1 \ge i \ge n $$

$$ X_1,...,X_n \text{ affine coord in } \mathbb{P}^n $$

Fix a $(x,y, t) \in \mathcal{X}$.

My question: Mumford uses terminology in this context that I don't understand. On pages 83 & 84 he talked about 'differentials of the equations (EQ) defining $\mathcal{X}$ restricted to tangent space $T_{(x,y,t),X \times Y, \{t\}} $, then ...'

I'm not familar with this terminology. What is a 'differential of an equation system' like (EQ) in this example?

Note, that we can regard $X_i$ as functions $\mathbb{A}^n \to \mathbb{C}$ from affine parts of $\mathbb{P}^n$, therefore compositions like $X_i \circ \sigma_t$ can be also regarded as locally defined functions $X \to \mathbb{C}$ and we can associate to differentaible functions a differential in usual way.

But here Mumford associates a differential to equations, instead to functions and I not understand it's meaning and construction. Is there a standard method in classical algebraic geometry to calculate a 'differential' of a equation system? In simple words: what is it?

1

There are 1 best solutions below

2
On BEST ANSWER

When talking about algebraic varieties it is common to speak interchangeably between "equations" that define a variety and the corresponding regular functions which vanish on the variety. So, when referring to the differential of an equation $f(x_1,\dots,x_n)=g(x_1,\dots,x_n)$, that would usually just mean the differential of $f(x_1,\dots,x_n)-g(x_1,\dots,x_n)$, the function whose vanishing is equivalent to that equation.