How to see that tangent cones really are cones?

971 Views Asked by At

I'm currently following an introductory course on algebraic geometry. When studying projective varieties, which we defined as $V(f)=\{ [x] | f[x]=0 \}$ given that f is a homogeneous polynomial, the following definition of tangent cones came up: If $\mathbb{C}P^n$ is a projective space and the projective variety $V(f) \subset \mathbb{C}P^n$, then we define the tangent cone with vertex $p \in V(f)$ as $C=V(\sum \frac{\delta^k f}{\delta x_{i_1}...\delta x_{i_k}}(p) x_{i_1}...x_{i_k})$ for every $x_{i_1}...x_{i_k}$ and k is the smallest integer for which there exists an $x_{i_1}...x_{i_k}$ so $ \frac{\delta^k f}{\delta x_{i_1}...\delta x_{i_k}}(p) \neq 0$. My question is this: Why is C a cone, ie. why will $pq \subset C$ if $q \in C$?

Thanks!

1

There are 1 best solutions below

3
On BEST ANSWER

Let's look at this on an affine patch; so assume that we're looking at $\{f=0\}$ in $\mathbb{A}^n$. Let's assume that the point we're looking at is $(0,0)$; if not, we do a linear change of coordinates. It is easy to see then that your definition of tangent cone is equivalent to looking at the zero set of the homogeneous part of $f$ of least degree. So, for example, if $f=y^2-x^2-x^3$, then the homogeneous part of least degree is $y^2-x^2$, and so the tangent cone is $y^2-x^2=0$; that is, $(y-x)(y+x)=0$.

Since it is defined by a homogeneous polynomial (let's say $G$), then if $x$ is in the tangent cone, we have that $G(\lambda x)=\lambda^dG(x)=0$, where $\lambda\in \mathbb{C}$.

This shows that it is a cone in the sense you're saying; now be careful, it is not a convex cone!