I have tried this first I took $x^2 + y^2 + 2xy$ on side and on the other side $x^2 + 4y^2 + 4xy$ and I complete square on both and this becomes $(x+y)^2 + (x + 2y)^2 = 1$ but this two line are not perpendicular to each other.
-
$4x^2+12xy+10y^2=2$, $(2x+3y)^2+y^2=2$. – Gerry Myerson May 09 '23 at 13:21
-
shift origin to centre of ellipse and rotate coordinate axes such that xy term cancels – Shlok Jain May 09 '23 at 13:28
-
Try to change to a new coordinate system, where your $"6xy"$ term will disappear by using change of coordinates such as $x = r\cos{\theta}$ and $y =r\sin{\theta}$. This will ease the completing-square process. Otherwise, consider calculus; $d/dx = 0$ and $d/dy = 0$, to evaluate the centre. – Dstarred May 09 '23 at 13:29
-
The axes of the ellipse lie on the eigenvectors of the associated matrix $\pmatrix{2&3\3&5}$. – Intelligenti pauca May 09 '23 at 13:30
-
But this two lines (2x + 3y) and y = 0 are not perpendicular to each other then how can it be axix of ellipse – Gourav Gupta May 09 '23 at 13:32
-
Ok I can rotate the axix but why can't this equation be converted into general form is it not possible – Gourav Gupta May 09 '23 at 13:34
-
@GouravGupta This lines need not be perpendicular. They don't represent axes of ellipse – Shlok Jain May 09 '23 at 13:35
-
Please don't be irritated by my question I am only in class 10th – Gourav Gupta May 09 '23 at 13:35
-
Why then they are not axis of ellipse – Gourav Gupta May 09 '23 at 13:41
2 Answers
Change the equation of an ellipse given in general form to what you say standard form, is frequently something tedious and still more tedious when the rectangular term is not null as in the case of the equation $$2x^2+6xy+5y^2=1$$ We show here a solution to motivated beginners.
A method is to find the angle $\theta$ in $\tan(2\theta)=\dfrac{6}{2-5}=-2$ (corresponding to $\dfrac{b}{a-c}$ in the general equation $ax^2+bxy+cy^2+dx+ey+f=0$) which gives $\theta\approx-0.553574358897$ and change of coordinates from $(x,y)$ to $(x',y')$ vía $$x=x'\cos\theta-y'\sin \theta\\y=x'\sin\theta+y'\cos\theta$$ which gives, (by confort we use $(x,y)$ instead of $(x',y')$) $$(2\cos^2\theta+5\sin^2\theta+6\sin\theta\cos\theta)x^2+6(\sin\theta\cos\theta+\cos^2\theta-6\sin^2\theta)xy+(2\sin^2\theta+5\cos^\theta-6\sin\theta\cos\theta)y^2=1$$ Calculation gives $$(0.14589803375)x^2+(6.85410196625)y^2=1$$ so we have the form $\dfrac{x^2}{a^2}+\dfrac{y^2}{b^2}=1$ where $a=2.61803398875$ and $b=0.38196601125$.
In the attached figure the black ellipse is the given one and the red ellipse is the new, after the changement of coordinates.
- 29,594
That is standard form that you got for the change of variables $(x+y,x+2y).$ In general, when faced with this kind of problem; ie, turn $$ax^2+bxy+cy^2=d$$ to standard form, just put $$x=\frac{p+q}{\sqrt{a}},\ y=\frac{p-q}{\sqrt{c}}.$$ This will always make the $xy$ term $\frac{b(p^2-q^2)}{\sqrt{ac}}.$ When you solve for $(p,q)$ in terms of $(x,y)$, you can always use matrices. This will always give orthogonal axes easily and algebraically.
So try the substitution $(x,y)=\left(\frac{p+q}{\sqrt{2}},\frac{p-q}{\sqrt{5}}\right).$
Substituting the above, we get
$$2(p^2+q^2)+\frac{6}{\sqrt{10}}(p^2-q^2)=1,$$ whereby we obtain $$\left(\frac{20+6\sqrt{10}}{10}\right)p^2+\left(\frac{20-6\sqrt{10}}{10}\right)q^2=1.$$ Now, this is your standard form. This is generally how you eliminate the $xy$ term without difficulty. Solving for $(x,y)$ in terms of $(p,q)$;
$$p+q=\sqrt{2}x\\ p-q=\sqrt{5}y\\ p=\frac{\sqrt{2}x+\sqrt{5}y}{2},\ q= \frac{\sqrt{2}x-\sqrt{5}y}{2}.$$
- 1,545
-
1
-
@Intelligentipauca (1) Yes, it is. The new ellipse has axes orthogonal to the $p$ and $q$-axes. (2) "Standard form" has nothing to do with orthogonality, just the form $X^2/a^2\pm Y^2/b^2=1$. Algebraically, orthogonal coordinates is an arbitrary choice, since an ellipse is a squashed, stretched and rotated circle, and a circle has no axes. To make them orthogonal, scale $p$ and $q$ appropriately with the conditions that the new scaling factors each have that the sum of their squares is 1. – Alexander Conrad May 10 '23 at 19:01
