0

Does $y = \dfrac{ax^2 + bx + c}{dx + e}$ have any lines of symmetry. If it does, what are they, and if not, how would one prove that it doesn't have any lines of symmetry?

(Please consider the general case. And please ignore trivial cases where the expression reduces to a quadratic expression in $x$, linear expression in $x$, or a constant.)

3 Answers3

1

I have written this solution independently of the solution of @Steven Gregory, with which it shares many common points . The differences are that

1) I have tried to concentrate on the main ideas, escaping computational details.

2) I am guided by the common general shape of the curves, an example of which I give below

3) I consider the exceptional case where the function is in fact affine.

Assuming $a \neq 0$ and $c \neq 0$, Euclidean division of the numerator by the denominator gives :

$$ax^2+bx+c=(Ax+B)(dx+e)+C \ \ \text{for certain real numbers} \ \ A,B,C.$$

Thus

$$f(x)=y=Ax+B+\dfrac{C}{dx+e} \ \ \text{with} \ \ A:=\frac{a}{d} \neq 0 \ \ (1)$$

Two cases:

  • Either: $C=0$ (which happens when $dx+e$ exactly divides $ax^2+bx+c$) in this case, the graphical representation of $f$ is a straight line which is its own symmetry axis and has an infinite number of axes of symmetry in the perpendicular direction...

  • Or $C\neq0$: we recognize here the equation of a hyperbola having a vertical asymptote with equation $x=-e/d$ and oblique (slant) asymptote with equation $y=Ax+B$. See graphics below. Thus the graphical representation has two axis of symmetry.

Explanation about the fact that the graphical representation of (1) is a hyperbola : let us write (1) under the form $(y-Ax-B)(cx+d)=C$ and make the affine change of variables $X=y-Ax-B, Y=cx+d$ (this does not modify the nature of the conical section) thus we obtain $XY=C$, which is one of the canonical forms for the equation of a hyperbola.

The example of $y=0.5x-1+\epsilon(x)$ with $\epsilon(x)=\frac{2}{x-1}\rightarrow 0 \ $ when $x \rightarrow \infty$, proving that $y=0.5x-1$ is the equation of the oblique asymptote (vertical asymptote at $x=1$):

enter image description here

Jean Marie
  • 81,803
  • I prefer your approach to mine. I would have used it if I had though of it. Except I don't think the OP has any idea what an affine space is. I went the way I did because I got a kick out of bisecting the vertical angles formed by the asymptotes. I also think it's cool that lines with slopes $\dfrac{a \pm \sqrt{a^2+d^2}}{d}$ are perpendicular to each other. – Steven Alexis Gregory Mar 20 '16 at 22:11
  • @Steven Gregory Thanks for your amability. Your solution had more than mine in the fact that you went through the explicit computations that the OP was asking for. – Jean Marie Mar 20 '16 at 23:26
1

$y=\dfrac{ax^2 + bx + c}{dx+e}$

can be rewritten as

$ax^2 - dxy + 0y^2 + bx + ey + c = 0$

which is known to be the equation of a conic section.

Comparing this to $Ax^2 + Bxy + Cy^2 + Dx + Ey + F$

we compute the descriminant to be $B^2 - 4AC = d^2 \gt 0$.

This implies that you have the equation of a hyperbola. Hence it will have two axis of symmetry.

addendum

$y=\dfrac{ax^2 + bx +c}{dx+e}$

\begin{array}{ccccccc} &&&& \dfrac adx &+& \dfrac{bd-ae}{d^2}\\ & &--&--&---&-&-----\\ dx + e&|&ax^2 &+& bx &+& c\\ & &ax^2 &+& \dfrac{ae}{d}x \\ & &---&-& -----\\ & && &\dfrac{bd-ae}{d}x &+& c \\ & && &\dfrac{bd-ae}{d}x &+& \dfrac{bde-ae^2}{d^2} \\ & && &-----&-&-------\\ & &&&&& \dfrac{cd^2 - bde + ae^2}{d^2} \\ \end{array}

We find $y = \dfrac adx + \dfrac{bd-ae}{d^2} + \dfrac{cd^2 - bde + ae^2}{d^2(dx+e)}$

Which admits to the vertical asymptote $dx + e = 0$ and the slant asymptote $adx-d^2y + (bd - ae) = 0$

The two asymptotes intersect at the point P = $\left(-\dfrac ed, \dfrac{bd - 2ae}{d^2}\right)$.

The change of coordinates \begin{array}{l} x = u - \dfrac ed\\ y = v + \dfrac{bd - 2ae}{d^2} \end{array}

translates the point $(x,y) = P$ to the point $(u,v) = (0,0)$

The asymptotes become $u = 0$ and $au -dv = 0$

The two axis of symmetry are the two lines that bisect the vertical angles formed by the two asymptotes. Their equations are described by

\begin{align} |u| &= \dfrac{|au-dv|}{\sqrt{a^2+d^2}} \\ au - dv &= \mp \sqrt{a^2 + d^2}\;u \\ (a \pm \sqrt{a^2 + d^2})u - dv &= 0 \\ v &= \dfrac{a \pm \sqrt{a^2 + d^2}}{d} u \end{align}

Translating back into $(x,y)$ coordinates, we get

\begin{align} y - \dfrac{bd - 2ae}{d^2} &= \dfrac{a \pm \sqrt{a^2 + d^2}}{d}(x + \dfrac ed) \\ y- \dfrac{bd - 2ae}{d^2} &= \dfrac{a \pm \sqrt{a^2 + d^2}}{d}x + \dfrac{e(a \pm \sqrt{a^2 + d^2})}{d^2}\\ y &= \dfrac{a \pm \sqrt{a^2 + d^2}}{d}x + \dfrac{-ea + bd \pm e\sqrt{a^2 + d^2})}{d^2} \end{align}

So the two axis of symmetry are $$y = \dfrac{a + \sqrt{a^2 + d^2}}{d}x + \dfrac{-ea + bd + e\sqrt{a^2 + d^2})}{d^2}$$

and

$$y = \dfrac{a - \sqrt{a^2 + d^2}}{d}x + \dfrac{-ea + bd - e\sqrt{a^2 + d^2})}{d^2}$$

  • Could you tell me what exactly the two axes of symmetry are, or point me to a reference? Thanks. –  Mar 20 '16 at 12:02
  • @KennyLJ - Why didn't you ask if that was what you wanted? You can find how to solve the general equation HERE in Wikipedia. – Steven Alexis Gregory Mar 20 '16 at 17:48
  • Thanks for the addendum. Another question -- How do you know that "The two axis of symmetry are the two lines that bisect the vertical angles formed by the two asymptotes"? –  Mar 21 '16 at 05:08
  • @KennyLJ - because the asymptotes are symmetrical about the axis of symmetry too. – Steven Alexis Gregory Mar 21 '16 at 12:19
-1

There does exist a line of symmetry when $a =1, b = c = d = 0, e = 1$, and the line is $x = 0$.

DeepSea
  • 77,651