An ellipsoid with semi-axes lengths of 1 0 , 1 5 , and 2 0 units, is to be placed in a rectangular box with a length of 3 3 , and a width of 2 5 . You want to determine the height of this box, such that when placing the ellipsoid in the box (and closing the lid) it is tangent to all of its six faces.
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.
The simplest solution of this problem depends on the fact that if an ellipsoid with semi-axes a , b , and c , is placed such that it is tangent to the three perpendicular planes, (and let the origin of the Cartesian Coordinates System be at their intersection point), then, the coordinates of the ellipsoid center C = ( C x , C y , C z ) satisfy,
C x 2 + C y 2 + C z 2 = a 2 + b 2 + c 2
Now, from symmetry, we know that after placing the ellipsoid in the box, with it tangent to all four vertical walls of the box, then the center of the ellipsoid x and y coordinates satisfy
C x = 2 L and C y = 2 W
And, hence, we can very easily now, compute the z -coordinate of the ellipsoid center C z , and it follows that the height of the box will be H = 2 C z , because the center of the ellipsoid is at the center of the box.
Cx^2+Cy^2+Cz^2=a^2+b^2+c^2 is true for the conjugate radiuses, which are not orthogonal -> is not true for a rectangular box.
I don't know what conjugate radiuses is ???
Log in to reply
Can you provide more information about: Cx^2+Cy^2+Cz^2=a^2+b^2+c^2 plz?
Log in to reply
Equation of ellipsoid: ( r − C ) T Q ( r − C ) = 1
where C = ( C x , C y , C z ) and Q = R D R T , with R being a rotation matrix and D is a diagonal matrix with diagonal entries = { 1 / a 2 , 1 / b 2 , 1 / c 2 }
Now place the ellipsoid such that it is tangent to the three planes in the first octant (simulating a rectangular box), and let the tangent point with the xy-plane be r 1 , then using the gradient of the ellipsoid, we have
Q ( r 1 − C ) = − α k , where α > 0 and k is the unit vector along the positive z-axis.
hence, r 1 − C = − α Q − 1 k ( ∗ )
substitute this into the equation of the ellipsoid, you obtain,
α 2 k T Q − 1 k = 1
from which α = k T Q − 1 k 1
finally taking the z-coordinate of equation (*)
C z = k T Q − 1 k
similar derivation leads to
C x = i T Q − 1 i
and
C y = j T Q − 1 j
hence,
C x 2 + C y 2 + C z 2 = i T Q − 1 i + j T Q − 1 j + k T Q − 1 k
but since Q = R D R T , then Q − 1 = R D − 1 R T , with D − 1 being a diagonal matrix of diagonal entries = { a 2 , b 2 , c 2 }
the right hand side of the above equation is the trace (the sum of the diagonal entries) of Q − 1 .
Now, since R T = R − 1 , then Q − 1 is related to D − 1 by a similarity transformation. And it is well-known that similar matrices have the same trace (see for example here ). Thus trace ( Q − 1 ) = trace ( D − 1 ) = a 2 + b 2 + c 2 , and we get our result.
Log in to reply
@Hosam Hajjir – Applying the same logic to an ellipse with a semi-axes a and b, tangent to two perpendicular lines, which cross at the origin of the coordinate system, then: Cx^2+Cy^2=a^2+b^2 . On the other hand (d1/2)^2+(d2/2)^2=a^2+b^2 , where d1 and d2 are conjugate diameters. For an ellipse, two diameters are conjugate if and only if the tangent line to the ellipse at an endpoint of one diameter is parallel to the other diameter. Each pair of conjugate diameters of an ellipse has a corresponding tangent parallelogram, which is rectangle only if conjugate diameters coincide with the axes or the ellipse is a circle. I found several online proofs (d1/2)^2+(d2/2)^2=a^2+b^2 , so Cx^2+Cy^2=a^2+b^2 is not correct.
I don't know much about conjugate diameters.
Log in to reply
Finally I found enough time to decipher your explanation and to realize it is absolutely correct. The trickiest part was the gradient of the ellipsoid. Thank you for the fantastic task and for clearing it to me!
Problem Loading...
Note Loading...
Set Loading...
The maximum and minimum values of u x + v y + w z , where u 2 + v 2 + w 2 = 1 , subject to the constraint a 2 x 2 + b 2 y 2 + c 2 z 2 = 1 , are ± a 2 u 2 + b 2 v 2 + c 2 w 2 . Thus the width of the ellipse a 2 x 2 + b 2 y 2 + c 2 z 2 = 1 in the direction of the unit vector ( u v w ) T is 2 a 2 u 2 + b 2 v 2 + c 2 w 2 . We need to find three mutually perpendicular unit vectors such that the widths of the given ellipse in the directions given by two of these vectors are 3 3 and 2 5 ; then the desired height of the box is the width of the ellipse in the direction of the third vector.
Thus we need to find mutually perpendicular unit vectors u and v such that 2 4 0 0 u 1 2 + 2 2 5 u 2 2 + 1 0 0 u 3 2 = 3 3 2 4 0 0 v 1 2 + 2 2 5 v 2 2 + 1 0 0 v 3 2 = 2 5 ( ⋆ ) If we consider the equations ( ⋆ ) , together with the equations u 1 2 + u 2 2 + u 3 2 = v 1 2 + v 2 2 + v 3 2 = 1 u 1 v 1 + u 2 v 2 + u 3 v 3 = 0 then there are many possible different solutions; for example we can obtain u = ⎝ ⎛ α β 1 1 − 3 β ⎠ ⎞ v = 2 0 1 ⎝ ⎛ 0 3 1 1 ⎠ ⎞ where α 2 = 9 2 5 4 1 4 β 2 = 1 8 5 0 0 5 1 1 Then the height of the box is 2 4 0 0 w 1 2 + 2 2 5 w 2 2 + 1 0 0 w 3 2 , where w = u × v . We calculate w = 5 3 7 0 1 ⎝ ⎛ 5 1 1 0 − 3 2 5 3 9 2 3 ⎠ ⎞ and the height of the box is therefore 1 1 8 6 = 3 4 . 4 3 8 4 .