Examples of classical conic duality

Example: The classic plane parabola

Let's work through how the classical duality map works for the classic plane parabola C, given by the equation y=x2 in R2.

First let's fix some points on C, say P1=(0,0), P2=(1,1) and P2=(2,4).

Computing tangent lines

Using standard calculus, determine the equation of the line Li tangent to the curve C at the point Pi, for each i.

Solution: By standard calculus, we have dydx=2x, and so using the point-slope formula yβˆ’y0=m(xβˆ’x0) we find that:


Next let's embed everything into P2(R), which from this point on we will simply call P2. Let R2β†ͺP2 be the standard embedding of the affine plane into the projective plane using the first affine coordinate chart, U1. In other words let it be the map that sends each point (a,b)∈R2 to the point [a:b:1]∈P2.

Projectivization

Write down the images of each point Pi, as well as the homogenized equations for C and the three tangent lines you found above.

Solution:

Under the given embedding, the points (0,0),(1,1),(2,4) correspond to the points [0:0:1],[1:1:1],[2:4:1] under the given embedding. As for the homogenizations of the different equations:

A note about common notation

It is common to use capital letters for coordinates of points (or for variables) in P2. In this case, the embedding R2β†ͺP2 corresponds to the coordinate equations x=XZ, Y=yz, Z=1. Using this convention and these equations, we would write the above homogenizations as X2βˆ’YZ=0, etc.


The zero sets of the homogenized equations of C (and each Li) determine projective varieties in P2, called the projective closures of the original varieties and denoted C― (and L―i, respectively).

Duals

  1. Using your homogeneous equations for the tangent lines L―i, write down the corresponding points L―iΛ‡ in the dual space PΛ‡2.
  2. Use the formula in Kathryn's summary to write down the equation for the dual C―ˇ of the curve C―.
  3. Verify that each point L―iΛ‡ lies on the dual conic C―ˇ.

Solution:

We first handle the duals of the tangent lines:

As for the dual conic, let's first recall the general duality result, snipped here from Kathryn's paper:

!800

Note that we are now using capital letters for our variables, and our equation for C― is X2βˆ’YZ=0. So for our curve we have a=1, e=βˆ’1, and b=c=d=f=0. If we use the variables U,V,W for the polynomials on the dual plane, then the dual curve should be given by U2βˆ’4VW=0.

It is now straightforward to verify that the points [0:1:0], [2:βˆ’1:βˆ’1] and [4:βˆ’1:βˆ’4] all satisfy this equation.


Example: The unit circle

The unit circle

Repeat the same steps above with the unit circle in R2, defined by the equation x2+y2=1. (Choose a few of your favorite points on the unit circle.)

Solution.

Let's use the following points:

Using implicit differentiation, we have 2x+2ydydx=0 and hence dydx=βˆ’xy. This lets us quickly compute the slopes of the tangent lines at each of our three points, and hence discover that the equations of the tangent lines at the three chosen points are:

Following our usual homogenization, the corresponding points in the dual projective space are [0:1:βˆ’1], [3:1:βˆ’2] and [1:βˆ’1:2], respectively.

Also, in standard format, the projectivization of our curve is given by X2+Y2βˆ’Z2=0, so a=c=1, f=βˆ’1 and b=d=e=0. The equation for the dual curve is then 4U2+4V2βˆ’4W2=0. It's now straightforward to verify the three dual points above do indeed lie on this curve.

Rather amazingly, this dual curve looks exactly like our original curve. In this case, it makes sense to say our curve is self dual.


Challenging question

Can you figure out how to deduce the equation for the dual conic? In other words, given a projective plane conic defined by ax2+bxy+cy2+dxz+eyz+fz2=0, can you find conditions for a projective line given by Ax+By+Cz=0 to be tangent to the conic? And if so, can you find the equation(s) on A, B and C that exactly describes all such tangent lines?