0

$$\frac{\left(\frac{xa^2}{a^2y^2+\ x^2}-p\right)^2}{a^2}+\left(\frac{ya^2}{a^2y^2+\ x^2}-q\right)^2=k^2$$

Could someone please convert this into standard form of equation

$$\frac{\left(X-H\right)^2}{A^2}+\frac{\left(Y-K\right)^2}{B^2}=1$$

I'm almost crying my brains out with all the algebra.

Background: This equation represents the inversion relative to the ellipse $\frac{x^2}{a^2}+{y^2}=1$ of a homothetic ellipse, i.e., of the ellipse $$ \frac{(x-p)^2}{a^2}+(y-q)^2=k^2 $$

amd
  • 53,693

3 Answers3

1

Hint: Rewrite the coefficients of $x$ and $y$ into a more recognizable form. That might suggest to you a way to finesse the problem instead of using brute-force algebra.


I’m going to give a general solution because I think that dropping the semi-minor $b$ from the equations obscures their symmetry.

Inverting the ellipse $$\frac{(x-p)^2}{a^2} + \frac{(y-q)^2}{b^2} = k^2$$ relative to $\frac{x^2}{a^2}+\frac{y^2}{b^2}=1$ yields, in standard form, the equation $$ \frac 1{a^2}\left(x-\frac p C\right)^2 + \frac 1{b^2}\left(y-\frac q C\right)^2 = \left(\frac k C\right)^2 $$ where $$ C = \frac{p^2}{a^2} + \frac{q^2}{b^2} - k^2 $$


Derivation: You can save yourself a lot of tedious algebraic manipulation by looking at inversion w/r a circle first.

To review, if $P$ is a point at distance $s$ from the origin, inversion w/r to a circle of radius $R$ centered at the origin maps it to the point $P'$ at a distance $s'$ along the (directed) line $\overrightarrow{OP}$ such that $ss'=R^2$, i.e., $s'=\frac{R^2}s$. This gives the inversion formula $$\eqalignno{ (x,y) \mapsto \left(\frac {R^2x}{x^2+y^2},\frac {R^2y}{x^2+y^2}\right) }$$ A circle of radius $k$ centered at $P$ intersects $\overrightarrow{OP}$ at $s+k$ and $s-k$, which are inverted to $\frac{R^2}{s+k}$ and $\frac{R^2}{s-k}$, respectively. Note that these values can be negative. The center of the circle’s image under inversion will be at their midpoint—not at the inversion of its center—with radius $k'$ equal to half the distance between them: $$\eqalign{\tag 1 k'=\frac 12\left(\frac{R^2}{s-k}-\frac{R^2}{s+k}\right) = \frac{R^2}{s^2-k^2}k \\ P' = \frac{R^2}{s+k} + k' = \frac{R^2}{s^2-k^2}s }$$ Let $C=\frac{s^2-k^2}{R^2}=\frac{p^2+q^2-k^2}{R^2}$ and $R=1$. Using the inversion formula we have for the equation of the circle’s image: $$ \left(\frac {x}{x^2+y^2}-p\right)^2+\left(\frac {y}{x^2+y^2}-q\right)^2=k^2 $$ but by $(1)$ this must be equivalent to $$\eqalign{\tag 2 \left(x-\frac p C\right)^2+\left(y-\frac q C\right)^2 = \left(\frac k C\right)^2 }$$ We can turn this into inversion relative to the ellipse $\frac{x^2}{a^2}+\frac{y^2}{b^2}=1$ by mapping to a coordinate system with $x$ and $y$ scaled by $a$ and $b$, respectively. The above considerations still hold, except that now the appropriate metric to use is $$ s^2 = \frac{x^2}{a^2}+\frac{y^2}{b^2} $$ so $$C=\left(\frac x a\right)^2+\left(\frac y b\right)^2-k^2$$ and the equation of the image of the ellipse $$\frac{(x-p)^2}{a^2}+\frac{(y-q)^2}{b^2}=k^2$$ is $$ \frac1{a^2}\left(\frac {x}{\frac{x^2}{a^2}+\frac{y^2}{b^2}}-p\right)^2+\frac1{b^2}\left(\frac {y}{\frac{x^2}{a^2}+\frac{y^2}{b^2}}-q\right)^2=k^2 $$ which by analogy to $(2)$ is $$ \frac1{a^2}\left(x-\frac p C\right)^2+\frac1{b^2}\left(y-\frac q C\right)^2 = \left(\frac k C\right)^2 $$

amd
  • 53,693
  • This was so helpful! Thanks, now i think I know how to do algebra again. Just one thing - I had a another question about the properties of ellipses, I googled it but I was pretty confused over the terms "curvatures" versus "radius of curvatures" and such. Was planning to put it on the comments but then I realized it would be better to create a new question altogether. I've the question here http://math.stackexchange.com/questions/1451959/curvature-of-ellipse. Could you please have a look? – soupynoodles Sep 26 '15 at 09:27
  • For dimensionsional reasons, the magic fudge factor might be better called $C^2$. – amd Sep 26 '15 at 17:48
  • okay, I love how you call it magic fudge :) – soupynoodles Sep 26 '15 at 18:00
1

$\begin{array}\\ k^2 &=\frac{\left(\frac{xa^2}{a^2y^2+\ x^2}-p\right)^2}{a^2}+\left(\frac{ya^2}{a^2y^2+\ x^2}-q\right)^2\\ k^2a^2 &=\left(\frac{xa^2}{a^2y^2+\ x^2}-p\right)^2+a^2\left(\frac{ya^2}{a^2y^2+\ x^2}-q\right)^2\\ k^2a^2(a^2y^2+\ x^2)^2 &=\left(xa^2-p(a^2y^2+\ x^2)\right)^2+a^2\left(ya^2-q(a^2y^2+\ x^2)\right)^2\\ &=x^2a^4-2pxa^2(a^2y^2+\ x^2)+p^2(a^2y^2+\ x^2)^2\\ &\quad+a^2(y^2a^4-2ya^2q(a^2y^2+ x^2)+q^2(a^2y^2+x^2)^2)\\ &=x^2a^4+y^2a^6\\ &\quad-(a^2y^2+x^2)(2pxa^2+2ya^4q)\\ &\quad+(p^2+a^2q^2)(a^2y^2+x^2)^2\\ &=a^4(x^2+y^2a^2)\\ &\quad-2a^2(a^2y^2+x^2)(px+ya^2q)\\ &\quad+(p^2+a^2q^2)(a^2y^2+x^2)^2\\ 0&=a^4(x^2+y^2a^2)\\ &\quad-2a^2(a^2y^2+x^2)(px+ya^2q)\\ &\quad+(p^2+a^2q^2-k^2a^2)(a^2y^2+x^2)^2\\ &=-a^2(a^2y^2+x^2)(2px+2ya^2q-a^2)\\ &\quad+(p^2+a^2q^2-k^2a^2)(a^2y^2+x^2)^2\\ &=(a^2y^2+x^2)(-a^2(2px+2ya^2q-a^2)\\ &\quad+(p^2+a^2q^2-k^2a^2)(a^2y^2+x^2))\\ 0&=-a^2(2px+2ya^2q-a^2)\\ &\quad+(p^2+a^2q^2-k^2a^2)(a^2y^2+x^2)\\ 0&=-a^2(2px+2ya^2q-a^2)\\ &\quad+C(a^2y^2+x^2)\\ &=Cx^2-2a^2px+a^4\\ &\quad+Ca^2y^2-2ya^4q\\ \end{array} $

And this is either an ellipse or hyperbola depending on the sign of $C$.

marty cohen
  • 107,799
  • It appears that you’ve introduced an extraneous solution, since the original equation represents an ellipse. Also, the OP wanted the equation in “standard form.” – amd Sep 25 '15 at 04:55
  • Well, I got it down to a form readily put into standard form. Also, with all that algebra, I could have certainly made a mistake. – marty cohen Sep 25 '15 at 04:57
  • @marty_cohen Thanks to both of you for trying to help. Yes, I'm converting this to standard form atm by completing the square, but since the "C" has three terms in it, I don't quite know what to do. How to simplify? – soupynoodles Sep 25 '15 at 05:00
  • Try different values of a, p, q, and k and see what you get. – marty cohen Sep 25 '15 at 05:21
  • 1
    well let me tell you what's going on here. Take an ellipse with minor axis $b=k$, and major axis $a$, so that its eccentricity is $\sqrt{\ 1-\frac{1}{a^2}}$. Then find its elliptic inverse across the ellipse with major axis $a$ and minor axis $b=1$. Now both the ellipse of inversion and the main ellipse I've talked about above are "homothetic" so the standard form has to be, by definition, an ellipse. I am trying various values of a, p, q, and k but it's not helping. Just gotta get that main thing into the form $$\frac{\left(X-H\right)^2}{A^2}+\frac{\left(Y-K\right)^2}{B^2}=1$$. idea? Thanks! – soupynoodles Sep 25 '15 at 05:32
  • That’s sort of what I was getting at in my hint: there’s an element of the form $\frac{u^2}{a^2}+\frac{v^2}{b^2}$ that keeps appearing in the expression. It seemed like it might make things easier to give that a name such as $\sigma(u,v)$ and substitute that into the equation. Anyway, what is it that you’re trying to do: simply prove that the image of an ellipse with the same eccentricity (and orientation) as the ellipse of inversion is also an ellipse, or actually find the equation of that image? – amd Sep 25 '15 at 08:33
  • Essentially, yes, I am looking for the equation of that ellipse. I did some algebra on my own, and here is what I've got - could you confirm if it's right? $$\left(x-\frac{a^2p}{C}\right)^2+a^2\left(y-\frac{a^2q}{C}\right)^2=\frac{\left(a^8+a^6q^2+a^4p^2\right)}{C^2}$$ Thanks. It's really untidy! – soupynoodles Sep 25 '15 at 13:44
  • 1
    @amd Sorry my mathjax is horrible! here it is again $$\left(x-\frac{a^2p}{C}\right)^2+a^2\left(y-\frac{a^2q}{C}\right)^2$$ = $$\frac{\left(a^8+a^6q^2+a^4p^2\right)}{C^2}$$ – soupynoodles Sep 25 '15 at 15:17
  • Looks nicer than mine. – marty cohen Sep 25 '15 at 18:03
  • I tested your solution by plotting an ellipse, its inversion, and your solution. Looks like you've got the inverted center correct, but the scale factor—the stuff on the r.h.s. is off. Try $k=p=q=1$: the size of the ellipse is very obviously wrong. – amd Sep 25 '15 at 20:37
  • @amd Yes i see! It's obviously a parabola I made, not an ellipse. I just completed the square but that's what I got! D'you know what's going on? – soupynoodles Sep 25 '15 at 20:43
0

To eliminate the $k^2$ term, just divide both sides of the equation by $k^2$. It will move from the right side into the denominators on the left.

Alex M.
  • 35,207
G. Allen
  • 184
  • Gotcha :) D'you know what to do when, after simplification, it looks like this? $$\frac{\left(a^2x^2+a^4y^2\right)}{\left(a^2y^2+x^2\right)^2k^2}+\frac{\left((p^2)+q^2a^2\right)}{a^2k^2}-\frac{\left(2px+2qa^2y\right)}{\left(a^2y^2+x^2\right)(k^2)}=1$$ – soupynoodles Sep 25 '15 at 03:45
  • I'm not sure your original equation represents an ellipse. If you let p = 0 and a = q = 1 then it becomes (x / – G. Allen Sep 25 '15 at 04:20
  • Sorry about the previous comment - the editor and I were having a disagreement. What I meant to say was: I'm not sure your original equation represents an ellipse. If you let p = 0 and a = q = 1 then it becomes

    $$(\frac{x}{x^2 + y^2})^2 + (\frac{y}{x^2 + y^2} - 1)^2 = k^2$$ $$(\frac{x}{x^2 + y^2})^2 + (\frac{y - x^2 - y^2}{x^2 + y^2})^2 = k^2$$ $$x^2 + (y - x^2 - y^2)^2 = (x^2 + y^2)^2k^2$$ $$x^2 + (y - x^2 - y^2)^2 = (x^4 + 2x^2y^2 + y^4)k^2$$

    I don't see how you can eliminate the fourth degree terms.

    – G. Allen Sep 25 '15 at 04:31
  • 1
    @G._allen well let me tell you what's going on here. Take an ellipse with minor axis $b = k$, and major axis $ka$, so that its eccentricity is $\sqrt{\ 1-\frac{1}{a^2}}$. Then find its elliptic inverse across the ellipse with major axis $a$ and minor axis $b=1$. What is the standard form of the new ellipse? – soupynoodles Sep 25 '15 at 05:09