5
$\begingroup$

In general, $$x^2+y^2+2gx+2fy+c=0$$ represents a circle with centre at $C(-g,-f)$. Equations of the form $$ax^2+2hxy+by^2=0$$ represents a pair of straight lines passing through origin. But the equation $2x^2+2y^2+xy=0$ represents just a point $(0,0)$.

When I came to know that it was not a pair of straight line equation I tried this: $$2x^2+2y^2+xy=2(x+y)^2-3xy=0$$ Using AM GM inequality, $(x+y)^2\ge 4xy$ $$2(x+y)^2\ge 8xy$$ $$3xy\ge 8xy$$ Which is only possible if $x=y=0$.

But is there an easy way to quickly check if the equation $ax^2+2hxy+by^2=0$ represents a point circle? After all, I used the above method only after knowing it's not a pair of straight lines.

$\endgroup$
  • 1
    $\begingroup$ I think you might find an answer trying to find out the connection of the radius and $c$ having a closer look at your first equation. (Have not tried it, it's just the thought coming to my mind first.) compare it with $\left(x+g\right)^2+\left(y+f\right)^2=r$. $\endgroup$ – Max Oct 22 '15 at 13:14
  • $\begingroup$ you can only use AM-GM if $x,y\geq 0$ $\endgroup$ – Dr. Sonnhard Graubner Oct 22 '15 at 13:15
  • 1
    $\begingroup$ I think $2x^2+2y^2+xy=0$ is not a point circle , since for a circle $xy$ term should be zero. That is general equation of point circle is $(x+g)^2+(y+f)^2=0$. So the equation $2x^2+2^2+xy=0$ are Pair of Imaginary straight lines with Real point of Intersection $(0,0)$ $\endgroup$ – Ekaveera Kumar Sharma Oct 22 '15 at 13:31
6
$\begingroup$

Take $ax^2+2hxy+by^2=0$. Multiply by $4a$ $(a\neq 0)$ and complete the square to obtain the equivalent equation $$(2ax+hy)^2+(4ab-h^2)y^2=0$$

If $4ab-h^2\gt 0$ then both terms on the left-hand side are non-negative, and must therefore be zero. If $4ab-h^2=0$ you get the single line $2ax+hy=0$ (the two lines coincide to give a degenerate case). And if $4ab-h^2\lt 0$ you get two lines by factoring the left-hand side as the difference of two squares.

$\endgroup$
  • 1
    $\begingroup$ It's perhaps worth saying that one can handle the case $a = 0$ by noting that the equation is unchanged under the replacement $(a, x) \leftrightarrow (b, y)$, so we can apply the same reasoning when $b \neq 0$. If $a = b = 0$, then the equation reduces to $h x y = 0$. If $h = 0$ the solution set consists of the entire plane, and if $h \neq 0$, the solution set is the union of the $x$- and $y$-axes. In each of these cases (except $a = b = h = 0$) the conclusion that the sign of $4 a b - h^2$ determines the character of the solution set still holds. $\endgroup$ – Travis Oct 22 '15 at 14:01
  • $\begingroup$ @Travis I was just thinking whether I should add that - so thank you! $\endgroup$ – Mark Bennet Oct 22 '15 at 14:02
  • $\begingroup$ You're welcome. And, anyway, the answer is quite nice (+1). $\endgroup$ – Travis Oct 22 '15 at 14:02
2
$\begingroup$

This is a special case of what is known as quadratic forms.

If you rewrite $ax^2+2hxy+by^2=0$ using matrices as $$\left[\matrix{x \\ y}\right]^T \left[\matrix{a & h \\ h & b}\right] \left[\matrix{x \\ y}\right] = 0 $$ you can perform a change of variables to the inner matrix's eigenvectors to get something on the form $$\left[\matrix{x' \\ y'}\right]^T \left[\matrix{d_1 & 0 \\ 0 & d_2}\right] \left[\matrix{x' \\ y'}\right] = 0 $$ which expands to $d_1{x'}^2 + d_2{y'}^2 = 0$

From this it is easy to see if it represents a single point, two intersecting lines or a single line just from the signs of $d_1$ and $d_2$ (or if one is zero).

If you instead had something non-zero on the right, you could instead get an ellipse, a hyperbola, or two parallell lines.

Quadratic forms also work in higher dimensions, but this answer is getting long as it is.
If you want, you can probably read more in a standard linear algebra textbook.

$\endgroup$
2
$\begingroup$

Note: @MarkBennet results gave me an idea of alternative approach.

Lets start off by assuming that $ax^{2}+2hxy+by^{2}=0$ represent a pair of straight lines (real/coincident/imaginary).

Substituting $\frac{y}{x}=m$ we have

$\Rightarrow bm^{2}+2hm+a=0$

$m=\frac{-h \pm \sqrt{h^{2}-ab}}{b}$

Plugging back $m=\frac{y}{x}$ we have two equations,

$$by=(-h + \sqrt{h^{2}-ab})x$$ and $$by=(-h - \sqrt{h^{2}-ab})x$$

These are the two lines represented by our equation $ax^{2}+2hxy+by^{2}=0$ , and as you can see we have an under root expression so we have two cases.

$(i)$ The lines are real and distinct if $h^{2}-ab>0$.

$(ii)$The lines are imaginary if $h^{2}-ab<0$.

$\endgroup$
  • $\begingroup$ @Mark Bennet has a better way. $\endgroup$ – Aditya Dev Oct 22 '15 at 15:26

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.