Consider the algebraic curve $\mathcal{C}$ given by $f(x,y)=0$, where $(x,y)\in\mathbb{C}^2$.

Suppose that the singular point of $f$ is $p=(x,y)=(0,0)$.

The blow-up of $p$ is given by $\{((x,y),[x_{1}:y_{1}])\in\mathbb{C}^{2}\times\mathbb{P}^1\}\rightarrow\mathbb{C}^2$. This produces the following two charts.

Chart 1

The blow-up of $p$ has the new coordinates $(x_{1},y_{1})$ given by $x_{1}=x$ and $y_{1}=\frac{y}{x}$.

The exceptional divisor, denoted as $E$, is given by $x_{1}=0$ and let $\widetilde{\mathcal{C}}$ denote the strict transform of $\mathcal{C}$.

Suppose that $E$ and $\widetilde{\mathcal{C}}$ intersect at $(x_{1},y_{1})=(0,\alpha)$, where $\alpha\neq 0$.

Chart 2

The blow-up of $p$ has the new coordinates $(x_{1}',y_{1}')$ given by $x_{1}'=\frac{x}{y}$ and $y_{1}'=y$.

The exceptional divisor, denoted as $E'$, is given by $y_{1}'=0$ and let $\widetilde{\mathcal{C}}'$ denote the strict transform of $\mathcal{C}$.

The result of Chart 1 implies that $E'$ and $\widetilde{\mathcal{C}}'$ should intersect at $(x_{1}',y_{1}')=(1/\alpha,0)$.


Why are the intersection points $(x_{1},y_{1})=(0,\alpha)$ and $(x_{1}',y_{1}')=(1/\alpha,0)$ equivalent? What is the geometric meaning of this?

Thanks, Jack.


Let me give an explicit example. Let $\mathcal{C}$ be a curve in $\mathbb{C}^2$ given by the vanishing of equation $f(x,y)=y^2-x^2-x^3$, a nodal cubic with a singularity at $(0,0)$. The blow up of $\mathcal{C}$ at point $(0,0)$ is defined as follows:

$Bl_{(0,0)} \mathbb{C}^2:= \{ ((x,y),[u:v]) \in \mathbb{C}^2 \times \mathbb{P}^1 \text{ } \vert \text{ } xv-yu=0 \} $, where $[u:v]$ are coordinates of $\mathbb{P}^1$ and $(x,y)$ are coordinates of $\mathbb{C}^2$.

Chart I. $u \neq 0$ $\implies u=1 \implies y=xv$

$\tilde{\mathcal{C}} \cup E$ in this chart is then given by equation $(xv)^2-x^2-x^3=0$ in $\mathbb{C}_{xv}^2$ where $x=0$ is the equation of the exceptional divisor $E$ in this chart and $v^2-1-x=0$ is the equation of $\tilde{\mathcal{C}}$ in this chart. Draw a picture - get a parabola intersecting a line. I.e. $\tilde{\mathcal{C}} \cap E = \{(0,1) , (0,-1) \}$. Emphasizing again: these are the intersection points in Chart I.

Chart II. $v \neq 0$ $\implies v=1 \implies x=yu$. In this chart, $\tilde{\mathcal{C}}$ is given by $1-u^2-yu^3=0$ in $\mathbb{C}_{uy}^2$ and the exceptional divisor is given by $y=0$. Get intersection points in Chart II: $\tilde{\mathcal{C}} \cap E = \{(1,0) , (-1,0) \}$.

Take, for example, the point $(0,1)$ in chart I and $(1,0)$ in chart II. They satisfy the criterion in your question. These 2 intersection points in different charts are "equivalent" or to be precise, they both represent the same point in the blow-up. Let's see that explicitly:

If we carefully go through what we did to the variables in Chart I, we see this: in Chart I, $u=1$, the point $(0,1)$ tells us $x=0, v=1$. Finally, $y=xv=0$. So, the point $(0,1)$ "back in the blow-up" is a point $((0,0),[1:1]) \in Bl_{(0,0)} \mathbb{C}^2$.

In Chart II, do the same. We get that $v=1$, $u=1, y=0$ and $x=0$. So again we get the point $((0,0),[1:1])$.

In this case, we can "see" the strict transform $\tilde{\mathcal{C}}$ in both charts. From my experience, we usually get something only in 1 chart. For example, if one blows up $V(x^2-y^3)$.

The simple thing is, both charts "see" a lot of the blow-up and sometimes they "see" overlapping things, that's normal, except very tedious to keep the track of.

  • $\begingroup$ Thank you. I overlooked the part where $[u,v]$ are the homogeneous in $((x,y),[u,v])$ so that $[u,v]=[1,a]=[1/a,1]$. $\endgroup$ – Jack Mar 13 '15 at 2:55

The blow-up of $\mathbb C^2$ in $(0,0)$ is not the entire set $\mathbb C^2\times \mathbb P^1$ but very imporantly, it is the incidence variety

$$\operatorname{Bl}_{(0,0)}(\mathbb C^2) = \{ (p,[q]) \mid p\in\mathbb C q \}\subseteq \mathbb C^2\times\mathbb P^1$$

The blow-up is then the projection $\pi:\operatorname{Bl}_{(0,0)}(\mathbb C^2)\to\mathbb C^2$.

The exceptional divisor of this blow-up is the projective line $E=\{(0,0)\}\times\mathbb P^1$.

So, you can consider those points $((x,y),[\tilde x:\tilde y])$ with $\tilde x\ne0$ and those with $\tilde y\ne0$. If you have $\tilde x\ne0$, you may scale to achieve $\tilde x=1$, then $(x,y)\in \mathbb C\cdot (1,\tilde y)$ so $x\tilde y = y$. For $x=0$ this implies $y=0$, so in this case your point is in $E$. If the point is not in $E$, then $\tilde y=\frac yx$ is defined by $(x,y)$, so the only point above $(x,y)$ in this case is $((x,y),[x:y])$.

This works analogously for the chart where $\tilde y\ne0$. If you have a point with $\tilde y\ne0$ and $\tilde x\ne0$, then still the only point above $(x,y)$ is the point $((x,y),[x:y])$.

I hope this helped? The geometric meaning is that you have $$\pi^{-1}(x,y) = \left\{\begin{array}{lcl}\{(0,0\}\times\mathbb P^1&;& (x,y)=(0,0)\\\{((x,y),[x:y]))\}&;&\text{otherwise}\end{array} \right.$$

so above every nonzero point of $\mathbb C^2$ there is a unique preimage - except for $(0,0)$ which has an entire copy of a $\mathbb P^1$ in its preimage. You may want to think of this $\mathbb P^1$ as parametrizing all the different possible tangents to a curve in the point $(0,0)$.

This way, if $\mathcal C$ has a singularity at $(0,0)$ which is due to the fact that it approaches the point with two distinct tangents, this blow-up is a strategy to separate these two approaches, pulling the curve slightly apart to remove the singularity. Mind you, though, a single blow-up will not remove every possible singularity.

  • $\begingroup$ Thank you very much for your help. $\endgroup$ – Jack Mar 13 '15 at 2:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.