An equation of an ellipsoid, when referred to its center as the coordinate system's origin, can be written as $$\tag{1} Ax^2+By^2+Cz^2+2Fyz+2Gxz+2Hxy+D=0$$

As I understand things, the values used for $x, y, z$ would be the Cartesian $x,y,z$ coordinates that describe the position of the points that make up the ellipsoid's surface.

If we were to multiply the coefficients in Eqn(1) by $(-1/D)$ we would then have $$\tag{2} ax^2+by^2+cz^2+2fyz+2gxz+2hxy=1$$

Would it be true that the $x,y,z$ values used in Eqn(2) are not the Cartesian coordinates of the points of the surface? Rather, the values to be used for $x,y,z$ are the components of the unit vector $l_i (l_1,l_2,l_3)$ that describes the orientation of the position vector that represents the position of a point on the ellipsoids surface? (I.e. the position vector is a radius vector $r$ drawn out from the origin in a direction pointing to the particular location on the surface of the ellipsoid).

Thus, we should write?: $$\tag{3} al_1^2+bl_2^2+cl_3^2+2fl_2l_3+2gl_1l_3+2hl_1l_2$$

  • 1
    $\begingroup$ Equations (1) and (2) are equivalent, so they are satisfied by the same $(x,y,z)$ on the surface of the ellipsoid. $\endgroup$ – GReyes Jan 3 '20 at 19:58
  • 1
    $\begingroup$ Equation $(2)$ is still an equation of the ellipsoid, it is not scaling it (turning it into a sphere) at all. Take a simple special case where $f, g, h=0$: you end up with $ax^2+by^2+cz^2=1$, i.e. $(x\sqrt{a})^2+(y\sqrt{b})^2+(z\sqrt{c})^2=1$. The point $(x\sqrt{a}, y\sqrt{b}, z\sqrt{c})$ now does lie on a sphere with radius $1$, so it looks like you would like to label $l_1=x\sqrt{a}, l_2=y\sqrt{b}, l_3=z\sqrt{c}$. $\endgroup$ – Stinking Bishop Jan 3 '20 at 20:05
  • $\begingroup$ To add to the previous comment, I am not sure if you call the "position vector" the vector $(x, y, z)$ scaled to magnitude $1$? It is, of course, $\left(\frac{x}{\sqrt{x^2+y^2+z^2}}, \frac{y}{\sqrt{x^2+y^2+z^2}}, \frac{z}{\sqrt{x^2+y^2+z^2}}\right)$, and is, in general, not the same as $(x\sqrt{a}, y\sqrt{b}, z\sqrt{c})$. $\endgroup$ – Stinking Bishop Jan 3 '20 at 20:17
  • 1
    $\begingroup$ Multiplying both sides of an equation by the same nonzero constant doesn’t change its solution set. $\endgroup$ – amd Jan 3 '20 at 20:33


$$ M = \left( \begin{array}{ccc} a & h & g \\ h & b & f \\ g & f & c \\ \end{array} \right) $$

an hermitian matrix and $X = (x,y,z)^{\dagger}$ we have

$$ X^{\dagger}MX = 1 $$

represents an ellipsoid as long as $\Lambda$ in

$$ M = Q^{-1}\Lambda Q $$

is a diagonal pos1tive matrix.

Here $Q$ is the normalized eigenvectors matrix associated to $M$. Making a coordinates change as $W = (w_1,w_2,w_3)^{\dagger} = Q X$ we have

$$ W^{\dagger}\Lambda W = 1 $$

This new coordinate system has for axes the eigenvectors of $M$ and in the $W$ coordinates the ellipsoid reads

$$ \lambda_1 w_1^2+\lambda_2 w_2^2+\lambda_3 w_3^2 = 1 $$


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.