0

Excuse the vagueness of this question, but how can you find the equation and distance for the diagonal of any given ellipse, that is, the line from the most-northwestern point to the most southeastern point?

The crude drawing below helps clarify: Imgur

Princee
  • 99
  • Ellipses don't have diagonals. Do you mean the major axis? Finding the equation of and the length of the major axis will depend on what information you are given about the ellipse. How is the ellipse given? – Gerry Myerson Apr 25 '14 at 00:13
  • @GerryMyerson I edited my question to further specify the "diagonal." – Princee Apr 25 '14 at 00:31
  • So I guess your ellipses are aligned with the major axis horizontal and the minor axis vertical. I would like to know how you define "most northwestern point". And I still want to know how the ellipse is given. If it's given as a drawing, about all you can do is take out a ruler and measure that "diagonal". – Gerry Myerson Apr 25 '14 at 00:38
  • @Princee What about my suggestion ? Did you read it ? What is your opinion ? – callculus42 Apr 26 '14 at 06:27
  • Do you mean, to be very precise, the two points at which the gradient of the tangent is $1$? – Mark Bennet Apr 26 '14 at 08:44

2 Answers2

2

Assume the ellipse is given by $$E:\quad{x^2\over a^2}+{y^2\over b^2}=1\ .\tag{1}$$ There are two ways to define the most-northwestern and most-southeastern points. The simpler idea consists in intersecting the $45^\circ$ lines $y=\pm x$ with the ellipse. I'm sure you can do this yourself.

But most probably you want the points where the tangent to $E$ is $45^\circ$ ascending. Intuitively they are the points where a $45^\circ$ line translated towards $E$ from far away first hits the ellipse. In order to determine these points we note that $E$ can be viewed as level line of the function $$f(x,y):={x^2\over a^2}+{y^2\over b^2}\ ,$$ and that at each point ${\bf z}\in E$ the gradient $$\nabla f(x,y)=\left({2x\over a^2},\>{2y\over b^2}\right)$$ is orthogonal to the tangent there. In the points ${\bf z}$ we are after we therefore have $\nabla f({\bf z})\perp(1,1)$. Tthis can be expressed by $\nabla f({\bf z})\cdot(1,1)=0$, or $${2x\over a^2}+{2y\over b^2}=0\ .\tag{2}$$ Equations $(1)$ and $(2)$ together determine the two points ${\bf z}_1$, ${\bf z}_2$ in question.

0

Edit: Modified example

for example you have an equation for an ellipse like this: $k=2x^2+8y^2+1$.

k is now 9.

$9=2x^2+8y^2+5x-3y+5$

To find the endings of the axes you can use the Lagrangian Multiplier Method. The Funktion is here $\mathfrak L=x^2+y^2+\lambda (-4+2x^2+8y^2+5x-3y+5)$

If you max/min this function you will get the four coordinates for the endings of the two axis.

greetings,

calculus

callculus42
  • 30,550
  • If you have the ellipse $2x^2+8y^2=8$, you'd be crazy to use Lagrange multipliers to find the ends of the axes. Just plug in $x=0$ and solve for $y$, then put in $y=0$ and solve for $x$, and you have the ends of the axes. But it seems that's not what OP wanted, anyway. – Gerry Myerson Apr 26 '14 at 05:51
  • @GerryMyerson You are right. But when the Ellipse is shifted it does´t work anymoere. And taking a ruler doesn´t really help. – callculus42 Apr 26 '14 at 06:23