MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Question. Let $C$ be a generic smooth curve of degree $d$ in $\mathbb{CP}^2$, and let $P$ be an arbitrary point away from this curve. How many lines are there through point $P$ that are tangent, or have tangency of order $k$ (for any $k$ between 3 and $d$) with $C$? Probably this can be done for small $d$ using the equation for $C$, but I would like to find out if there is a formula for general $d$.

share|cite|improve this question
up vote 3 down vote accepted

You have some polynomial $f(x,y,z)$. A line through the point $(1:0:0)$ can be paramaterized by a map from $\mathbb P^1: (u:v) \to (u:av:bv)$ for some constant $a$ and $b$. $f$ restricts to a degree $d$ polynomial in $u$ and $v$. Since it has no roots where $v=0$, set $v=1$. You now have a univariate polynomial such that the $k$th coefficient is a degree $k$ polynomial in $a$ and $b$. The discriminant of this polynomial determines whether it has a double root, which is the same as if the line is tangent. The discriminant is a degree $d(d-1)$ polynomial in $a$ and $b$.

Thus, the generic number of roots is $d(d-1)$. This is similarly the generic number of tangent lines. The generic number of inflection points or higher is zero, as this would require a multiple root of the discriminant.

To check that there is not some extra constraint on the discriminant polynomial that forces a multiple root, we can just find an example that does not have a multiple root. If $f(x,y,z)=x^d+xy^{d-1}+z^d$, then $f(u)=u^d+ua^{d-1}+b^d$, whose discriminant, which is $a^{d^2-d}+b^{d^2-d}$ up to some constant factors, has no multiple roots.

share|cite|improve this answer
@Will thanks for your answer – user23802 May 24 '12 at 6:13
You can also find the tangency points as intersections with the polar curve $\partial f/\partial x=0$. This is an alternative way to see that there are $d(d-1)$ such points. If the base point of the pencil is $[a:b:c]$, you need the polar to be $a \partial f/\partial x + b \partial f/\partial y +c \partial f/\partial z=0$. For non generic curves the amount by which $d(d-1)$ is bigger than the actual number can be controlled (depending on higher order tangencies and singularities). Look for "Plücker formulas". – quim May 24 '12 at 9:34

Let $\tilde\pi:\mathbb{PC}^2\setminus\{P\}\to\mathbb{PC}$ be the projection from $P$, which restricts to a surjective morphism $\pi:C\to\mathbb{PC}$ of degree $d$. You are asking for the number of points of ramification of this morphism with multiplicity, or more precisely the degree of the ramification divisor $R$. By Hurwitz' theorem, it can be computed as $$\deg(R) = 2g(C)-2 - d(2g(\mathbb{PC}) - 2)=2\binom{d-1}{2} + 2(d-1) = d(d-1).$$

share|cite|improve this answer
@Jesko thanks for your answer. – user23802 May 24 '12 at 6:15
A technical note: There is of course no morphism from $\mathbb {CP}^2$ to $\mathbb {CP}$. You mean to say that $\bar{\pi}$ is a morphism from $\mathbb {CP}^2-\{P\}$. – Will Sawin May 24 '12 at 6:16
@Will: Thanks, I corrected it. – Jesko Hüttenhain May 24 '12 at 6:46

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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