EXERCISE: Let $h\in\mathbb{R}$ and consider the conic section $C:x^2+2hxy+y^2-h-4$. Are there any values for which $C$ is a parabola? What kind of conic section is $C$ for $h=1$? And for $h=-1$?

SOLUTION: I investigate the determinants of the matrix of the quadratic equation $B$ and that of the matrix of the quadratic form $A$: \begin{equation} det(B)=\begin{vmatrix} 1 & h & 0 \\ h & 1& 0\\0&0&-h-4 \end{vmatrix} = (-h-4)\begin{vmatrix}1&h\\h&1\end{vmatrix}=(-h-4)(1-h^2)=h^3+4h^2-h-4 \end{equation} \begin{equation} det(A)=\begin{vmatrix}1&h\\h&1\end{vmatrix}=1-h^2 \end{equation} In order for $C$ to be a parabola, $det(B)$ needs to be non-zero, hence $h\neq\pm1\wedge h\neq-4$ (as $h^3+4h^2-h-4=(h+1)(h-1)(h+4)$), and $det(A)$ needs to be zero. Since the second condition implies $h=\pm1$, $C$ can never be a parabola.

Now, if $h=\pm1$, $B$ is singular, so that $C$ is a degenerate conic. In order to claffify what type of degenerate conic I have to inspect the determinant of $A$ but, by substituting $h=\pm1$ I get the zero. I have to conclude that for such values $C$ is not a conic or what? Have I made some mistakes?

Thank you all for you help.

  • $\begingroup$ you only need to look at the coefficient matrix of quadratic terms, that $\left[\begin{array}{cc}1 & h\\ h &1\end{array}\right]$, when is it positive definite or have zero eigenvalues? $\endgroup$ – m-agag2016 Dec 12 '16 at 10:50
  • $\begingroup$ I do not know what a positive definite matrix is, but I looked online right now and I found that, through Sylvester's criterion, I can say that a matrix is positive definite if all of the leading principal minors are positive. In this case I have the principal minor of order one positive $(1>0)$ while that of order 2 is $1-h^2$ which is strickly greater than zero only if $h<0$. Anyway, I cannot see how this helps :( $\endgroup$ – M.Giacchello Dec 12 '16 at 11:03
  • $\begingroup$ For $h=1$, we have $(x+y)^2=5$. For $h=-1$, we have $(x-y)^2=3$. You may want to see here. $\endgroup$ – mathlove Dec 12 '16 at 11:04
  • $\begingroup$ en.m.wikipedia.org/wiki/… $\endgroup$ – lab bhattacharjee Dec 12 '16 at 12:01

I found that both $det(B)=det(A)=0$ hence, as suggested by @mathlove, I have to inspect the determinant of the matrix $C$, which is: \begin{equation}det(C)=\begin{vmatrix}a_{11}&a_{13}\\a_{13}&a_{33}\end{vmatrix}+\begin{vmatrix}a_{22}&a_{23}\\a_{23}&a_{33}\end{vmatrix}\end{equation} So, for $h=1$ I get \begin{equation}det(C)=\begin{vmatrix}1&0\\0&-5\end{vmatrix}+\begin{vmatrix}1&0\\0&-5\end{vmatrix}=-10<0\end{equation} and thus two real-parallel lines.

For $h=-1$ \begin{equation}det(C)=\begin{vmatrix}1&0\\0&-3\end{vmatrix}+\begin{vmatrix}1&0\\0&-3\end{vmatrix}=-6<0\end{equation} that is, again, a couple of parallel lines.

NOTE: I answered my own question just to close the topic. Thank you very much to everyone and especially to mathlove!


A general conic in the affine plane has equation

$$ax^2 + by^2 +2hxy + 2gx+2fy+c = 0$$

but they're better though of as curves in the projective plane. In homogeneous coordinates:

$$ax^2 + by^2 +2hxy + 2gxz+2fyz+cz^2 = 0$$

The line at infinity has equation $z=0$, and the intersections with the line at infinity have

$$ax^2 + by^2 +2hxy = 0$$

Hyperbolae meet the line at infinity twice. Parabolae are tangent to the line at infinity. Ellipses miss the line at infinity. The number or roots of $ax^2 + by^2 +2hxy = 0$ classifies the conic.

Assuming that $a \neq 0$, $$\frac{x}{y} = \frac{-h\pm\sqrt{h^2-ab}}{a}$$ Assuming that $b \neq 0$, $$\frac{y}{x} = \frac{-h\pm\sqrt{h^2-ab}}{b}$$

We have a hyperbola is $h^2-ab > 0$, a parabola if $h^2-ab=0$ and an ellipse if $h^2-ab<0$.


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.