Ellipsoid in the corner

Geometry Level pending

An ellipsoid is pushed against the corner of the first octant (where x 0 , y 0 x \ge 0 , y \ge 0 and z 0 z \ge 0 ). The ellipsoid cannot penetrate the coordinate planes (the x y xy , x z xz , and y z yz planes), and can only be tangent to them. It is oriented such that its first semi-axis which has a length of 3 3 is along the vector ( 4 , 1 , 8 ) (4, 1, 8) , and its second semi-axis which is of length 4 4 is along the vector ( 4 , 8 , 1 ) (4, -8, -1 ) , and the third semi-axis, of length 7 7 is along the vector ( 7 , 4 , 4 ) (7, 4, -4) .

There are three tangency points with the three coordinate planes. If ( 0 , a , b ) (0, a, b) is the tangency point with the y z yz plane, then find a + b a + b .


The answer is 9.387.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

1 solution

Hosam Hajjir
Feb 8, 2019

The parametric equation (vector equation) of points on the ellipsoid surface is

r ( θ , ϕ ) = r C + V u ( θ , ϕ ) \mathbf{r}(\theta, \phi) = \mathbf{r}_C + V \mathbf{u}(\theta,\phi)

where r C = ( x C , y C , z C ) \mathbf{r}_C = (x_C , y_C , z_C) is the position vector of the center of the ellipsoid. And the 3 × 3 3 \times 3 matrix V V is given by

V = [ v 1 , v 2 , v 3 ] = R D V = [ v_1, v_2, v_3 ] = \mathbf{R D} , with R \mathbf{R} being an orthogonal matrix whose columns are the unit vectors of three axes of the ellipsoid, and D \mathbf{D} is a diagonal matrix of the semi-axes lengths. Finally, the unit vector u ( θ , ϕ ) \mathbf{u}(\theta,\phi) is specified as

u ( θ , ϕ ) = ( sin θ cos ϕ , sin θ sin ϕ , cos θ ) \mathbf{u}(\theta,\phi) = ( \sin \theta \cos \phi, \sin \theta \sin \phi, \cos \theta )

At the three tangency points, one of the coordinates vanishes, and the same coordinate of the derivatives of r ( θ , ϕ ) \mathbf{r}(\theta, \phi)

with respect to θ \theta and to ϕ \phi vanishes as well.

Let's take the tangent point with the x y xy plane. We know that at that point we'll have,

z C + v 1 z sin θ 1 cos ϕ 1 + v 2 z sin θ 1 sin ϕ 1 + v 3 z cos θ 1 = 0 z_C + v_{1z} \sin \theta_1 \cos \phi_1 + v_{2z} \sin \theta_1 \sin \phi_1 + v_{3z} \cos \theta_1 = 0

In addition, by differentiating r \vec{r} with respect to θ \theta and ϕ \phi and setting the z z -component to zero,

we obtain two more equations,

v 1 z cos θ 1 cos ϕ 1 + v 2 z cos θ 1 sin ϕ 1 v 3 z sin θ 1 = 0 v_{1z} \cos \theta_1 \cos \phi_1 + v_{2z} \cos \theta_1 \sin \phi_1 - v_{3z} \sin \theta_1 = 0

and

v 1 z sin θ 1 sin ϕ 1 + v 2 z sin θ 1 cos ϕ 1 = 0 -v_{1z} \sin \theta_1 \sin \phi_1 + v_{2z} \sin \theta_1 \cos \phi_1 = 0

Now in the last equation, sin θ 1 = 0 \sin \theta_1 = 0 is extraneous and can be ignored, and therefore we can divide by it, resulting in

v 1 z sin ϕ 1 + v 2 z cos ϕ 1 = 0 -v_{1z} \sin \phi_1 + v_{2z} \cos \phi_1 = 0

And this can be solved for ϕ 1 \phi_1 , resulting in,

cos ϕ 1 = v 1 z v 1 z 2 + v 2 z 2 \cos \phi_1 = - \dfrac{v_{1z}} { \sqrt{{v_{1z}}^2 + {v_{2z}}^2 }} and

sin ϕ 1 = v 2 z v 1 z 2 + v 2 z 2 \sin \phi_1 = - \dfrac{v_{2z}} { \sqrt{{v_{1z}}^2 + {v_{2z}}^2 }}

Plugging in the angle ϕ 1 \phi_1 in the previous equation, we can now solve for θ 1 \theta_1 , in a similar fashion, and

the result is that,

cos θ 1 = v 3 z v 1 z 2 + v 2 z 2 + v 3 z 2 \cos \theta_1 = -\dfrac{v_{3z}}{ \sqrt{{v_{1z}}^2 + {v_{2z}}^2 + {v_{3z}}^2}}

sin θ 1 = v 1 z 2 + v 2 z 2 v 1 z 2 + v 2 z 2 + v 3 z 2 \sin \theta_1 = \dfrac{ \sqrt{{v_{1z}}^2 + {v_{2z}}^2 }}{ \sqrt{{v_{1z}}^2 + {v_{2z}}^2 + {v_{3z}}^2}}

The choice of signs for cos ϕ 1 \cos \phi_1 , sin ϕ 1 \sin \phi_1 , cos θ 1 \cos \theta_1 and sin θ 1 \sin \theta_1 guarantees that θ 1 [ 0 , π ] \theta_1 \in [0, \pi ] on the one hand, and also that the z z -coordinate of the tangency point (which is 0) is as far as possible from z C z_C (which is positive) for any values of v 1 z v_{1z} , v 2 z v_{2z} and v 3 z v_{3z} .

Plugging both θ 1 \theta_1 and ϕ 1 \phi_1 into the first equation, results in

z C = v 1 z 2 + v 2 z 2 + v 3 z 2 z_C = \sqrt{{v_{1z}}^2 + {v_{2z}}^2 + {v_{3z}}^2}

Similarly, we have,

x C = v 1 x 2 + v 2 x 2 + v 3 x 2 x_C = \sqrt{{v_{1x}}^2 + {v_{2x}}^2 + {v_{3x}}^2}

and

y C = v 1 y 2 + v 2 y 2 + v 3 y 2 y_C = \sqrt{{v_{1y}}^2 + {v_{2y}}^2 + {v_{3y}}^2}

So far we have found the center of the ellipsoid, and expressions for the angles θ \theta and ϕ \phi at the tangency point.

Now for the tangency point at the y z yz plane, we have the following,

cos ϕ 3 = v 1 x v 1 x 2 + v 2 x 2 \cos \phi_3 = - \dfrac{v_{1x}} { \sqrt{{v_{1x}}^2 + {v_{2x}}^2 }} and

sin ϕ 3 = v 2 x v 1 x 2 + v 2 x 2 \sin \phi_3 = -\dfrac{v_{2x}} { \sqrt{{v_{1x}}^2 + {v_{2x}}^2 }}

and

cos θ 3 = v 3 x v 1 x 2 + v 2 x 2 + v 3 x 2 \cos \theta_3 = - \dfrac{v_{3x}}{ \sqrt{{v_{1x}}^2 + {v_{2x}}^2 + {v_{3x}}^2}}

sin θ 3 = v 1 x 2 + v 2 x 2 v 1 x 2 + v 2 x 2 + v 3 x 2 \sin \theta_3 = \dfrac{ \sqrt{{v_{1x}}^2 + {v_{2x}}^2 }}{ \sqrt{{v_{1x}}^2 + {v_{2x}}^2 + {v_{3x}}^2}}

Using these values, we can find r ( θ 3 , ϕ 3 ) \mathbf{r}(\theta_3, \phi_3 ) using the parametric representation of the ellipsoid. And the result is r = ( 0 , 2.855166 , 6.531751 ) \mathbf{r}^* = ( 0, 2.855166 , 6.531751 ) , making the answer ( 2.855166 + 6.531751 ) = 9.386917 (2.855166+6.531751) = \boxed{9.386917}

Another approach to this problem is to consider the algebraic equation of the ellipsoid instead of the vector equation.

The algebraic equation of ellipsoid is given by,

( r r C ) T R D R T ( r r C ) = 1 (\mathbf{r} - \mathbf{r}_C)^T \mathbf{R D R}^T (\mathbf{r} - \mathbf{r}_C) = 1

where the matrix R \mathbf{R} is as before, but the diagonal matrix D = diag { 1 a 2 , 1 b 2 , 1 c 2 } \mathbf{D} = \text{diag} \{ \dfrac{1}{a^2}, \dfrac{1}{b^2},\dfrac{1}{c^2} \} , where a , b a, b and c c are the semi-axes lengths. The gradient of the above expression (the normal to the surface) is given by,

= 2 R D R T ( r r C ) \nabla = 2 \mathbf{R D R}^T (\mathbf{r} - \mathbf{r}_C)

Let r 1 \mathbf{r}_1 be the tangency point with the x y xy plane, then at that point the gradient is pointing in the opposite direction of k \mathbf{k} (the unit vector along the positive z z axis ). Hence, there exist a positive constant α \alpha , such that,

R D R T ( r 1 r C ) = α k \mathbf{R D R}^T (\mathbf{r}_1 - \mathbf{r}_C)= - \alpha \mathbf{k}

and this implies that,

r 1 r C = α R D 1 R T k ( ) \mathbf{r}_1 - \mathbf{r}_C= - \alpha \mathbf{R D}^{-1} \mathbf{R}^T \mathbf{k} \hspace{24pt}(*)

Substituting this expression into the algebraic equation of the ellipsoid, and solving for α \alpha , we get,

α = 1 k T R D 1 R T k \alpha = \dfrac{1}{\sqrt{ \mathbf{k}^T \mathbf{R D}^{-1} \mathbf{R}^T \mathbf{k} }}

Since the z z coordinate of r 1 = 0 \mathbf{r}_1 = 0 , it follows that, by pre-multiplying both sides of equation ( ) (*) by k T \mathbf{k}^T ,

z C = α k T R D 1 R T k = k T R D 1 R T k z_C = \alpha \mathbf{k}^T \mathbf{R D}^{-1} \mathbf{R}^T \mathbf{k} = \sqrt { \mathbf{k}^T \mathbf{R D}^{-1} \mathbf{R}^T \mathbf{k} }

and similarly,

x C = i T R D 1 R T i x_C = \sqrt{ \mathbf{i}^T \mathbf{R D}^{-1} \mathbf{R}^T \mathbf{i} }

y C = j T R D 1 R T j y_C = \sqrt{ \mathbf{j}^T \mathbf{R D}^{-1} \mathbf{R}^T \mathbf{j} }

where i \mathbf{i} and j \mathbf{j} are the unit vector along the positive x x and y y axes. Thus we have determined r C \mathbf{r}_C

And using equation ( ) (*) and its variations, we can obtain expressions for the three tangent points r 1 \mathbf{r}_1 with the x y xy plane , r 2 \mathbf{r}_2 , with the x z xz plane, and r 3 \mathbf{r}_3 with the y z yz plane, and they are as follows,

r 1 = r C R D 1 R T k k T R D 1 R T k \mathbf{r}_1 = \mathbf{r}_C - \dfrac{ \mathbf{R D}^{-1} \mathbf{R}^T \mathbf{k} }{ \sqrt{ \mathbf{k}^T \mathbf{R D}^{-1} \mathbf{R}^T \mathbf{k} } }

r 2 = r C R D 1 R T j j T R D 1 R T j \mathbf{r}_2 = \mathbf{r}_C - \dfrac{ \mathbf{R D}^{-1} \mathbf{R}^T \mathbf{j} }{ \sqrt{ \mathbf{j}^T \mathbf{R D}^{-1} \mathbf{R}^T \mathbf{j} } }

r 3 = r C R D 1 R T i i T R D 1 R T i \mathbf{r}_3 = \mathbf{r}_C - \dfrac{ \mathbf{R D}^{-1} \mathbf{R}^T \mathbf{i} }{ \sqrt{ \mathbf{i}^T \mathbf{R D}^{-1} \mathbf{R}^T \mathbf{i} } }

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...