CHAPTER 11

Conics and Polar Coordinates

11.1. Quadratic Relations

We will see that a curve dened by a quadratic relation between the variables x y is one of these three curves: a) parabola, b) ellipse, c) hyperbola. There are other possibilities, considered degenerate. For example the graph of the equation x2 y2 a we know to be a circle, if a 0. But if a 0, the graph is just the point 0 0, and if a 0, there is no graph. Similarly the equation x2 y2 a describes a hyperbola if a 0, but if a 0, we get the two lines x y. First we list the standard forms of the basic curves. These are standard in the sense that any other curve given by a quadratic equation is obtained from one of these by moving the curve in the plane by translating and/or rotating. The Parabola. The standard form is one of these: (11.1) y ax2 x ay2

The sign of a determines the orientation of the parabola. We have these four possibilities: y ax2 a 0 Figure 11.1 y ax2 a 0 Figure 11.2 x ay2 a 0 Figure 11.3 x ay2 a 0 Figure 11.4

Figure 11.1 y

Figure 11.2 y

a 1 x a

1 x

157

Chapter 11

Conics and Polar Coordinates

158

Figure 11.3 y 1 a x

Figure 11.4 y

1 a x

The magnitude of a determines the spread of the parabola: for a very small, the curve is narrow, and as a gets large, the parabola broadens. The origin is the vertex of the parabola. In the rst two cases, the y-axis is the axis of the parabola, in the second two cases it is the x-axis. The parabola is symmetric about its axis. The Ellipse The standard form is (11.2) x2 a2

y2 b2

The values x can take lie between a and a and the values y can take lie between b and b. If a b (as shown in gure 11.5), the major axis of the ellipse is the x-axis, the minor axis is the y-axis and the points a 0 are its vertices.
Figure 11.5 Figure 11.6 b b

a b

a b

If a b (as shown in gure 11.6), the major axis of the ellipse is the y-axis, x 0 is the minor axis, and the points 0 b are its vertices. Of course, if a b, the curve is the circle of radius a, and there are no special vertices or axes. The Hyperbola. The standard form is one of these: (11.3) x2 a2 y b2
2

y2 b2

x a2

corresponding to gures 11.7 and 11.8.

11.1
Figure 11.7

Quadratic Relations

159

Figure 11.8

The x-axis is the axis of the rst hyperbola. The points a 0 are the vertices of the hyperbola; for x between these values, there corresponds no point on the curve. We similarly dene the axis and vertices of the hyperbola of gure 11.8. The lines (11.4) y

bx a

are the asymptotes of the hyperbola, in the sense that, as x , the curve gets closer and closer to these lines. We see this by dividing the dening equation by x2 , and consider what happens as x . For example, using the rst equation, we get (11.5) 1 a2 y b12 x2
2

1 x2

Figure 11.9

b c a

which we can rewrite as (11.6) a 2 y2 b 2 x2 1 1 x2

so that, as x gets large, the hyperbola approaches the graph of (11.7) which amounts to the two equations y a 2 y2 b 2 x2 1

ax. Figure 11.9 shows these asymptotes.

Chapter 11

Conics and Polar Coordinates

160

Now, the general quadratic relation between x and y is (11.8) Ax2 By2 Cxy Dx Ey F 0

If C 0, then by completing the square in both x and y we are led to an equation which looks much like one of the standard forms, but with the center removed to a new point x0 y0 . If C 0, the situation is more difcult: a rotation of the gure is also required to get it into standard form. We will discuss this no further, and consider only the case C 0. First, some examples:
70 50 30 10 0 2
5

Example 11.1 Graph the curve 3x2 30x y 73 We have to complete the square in x. We get (11.9) 3x2 10x 25 y 73 75 3 x 5 2

0 0

2
6 8 10

which gives the standard form (11.10) y2

Figure 11.10

Example 11.2 Graph the curve 9x2 4y2 18x 16y 11 Completing the squares: (11.11) 9x
2

0 36

4 2

2x 1 4y 4y 4
2

11 9 16

6 4 2

(11.12)

12 y 22
22 32

2 4 6

Figure 11.11
4 2

Example 11.3 Graph the curve 5x2 y2 30x 4y 46 the squares:

2 4 6
3

0 Completing 5

6 4 2 0 2 4 6 8

(11.13)

5x2 6x 9 y2 4y 4
y 22

46 45 4

(11.14)

5 2

x 32

Figure 11.12

Proposition 11.1 The equation (11.15) Ax2 By2 Dx Ey F 0

can be put into one of the following forms by completing the square: a) (parabola):y y0 Ax x0 2 if B 0 The vertex of the parabola is at x0 y0 , and the axis is the line x x0 . b) (parabola): x x0 Cy y0 2 if A 0 The vertex of the parabola is at x0 y0 , and the axis is the line y y0 .

11.2
c) (ellipse)
x

Eccentricity and Foci

161

1 if A and B are of the same sign The center of the ellipse is at a2 b2 x0 y0 , and its axes are the lines x x0 y y0 . 2 y y0 2 x x0 2 y y0 2 d) (hyperbola): b2 1 or x 2x0 1 if A and B are of different 2 2 a b a signs. The center of the hyperbola is x0 y0 , and its axes are the lines x x0 y y0 . e If both A and B are zero, the curve is a line. The following degenerate cases may also result: (11.16) Ax x02 By y02 Ax x02 By y02 0 : no graph or just the point x0 y0

x02 y y02

(11.17)

0 :

two lines crossing at x0 y0

Example 11.4 Finally, just to illustrate the situation of a quadratic whose coefcient of xy is nonzero, we consider the curve xy 1 0. This curve is symmetric about the lines y x, and has the asymptotes x 0 y 0. This appears to be a hyperbola with major axis the line x y. In fact, if we make the linear change of variables x u v y u v, this becomes the curve u2 v2 1 in the new variables. (This change of variables represents a rotation by 45 , with a slight change of scale.)

11.2. Eccentricity and Foci

These curves are called the conic sections because they can be visualized as the intersection of a cone with a plane. We shall now consider another denition, dating from the ancient Greeks, which leads to important properties of the conics. Fix a point F and a line L in the plane such that L does not go through F. Pick a positive number e. We consider the locus C of all points X in the plane such that (11.18) XF e XL

where XY means the distance from X to Y . e is the eccentricity of C; F the focus and L the directrix. Note that the curve C is symmetric about the line through the focus and perpendicular to the directrix. This is the axis of the curve. There is one point between F and L on C which is on this axis; this point is the vertex of C. We now show that if e 1, C is a parabola, if e 1, C is an ellipse and if e 1, C is a hyperbola. Lets take the axis of C to be the x axis, and place the vertex at the origin, O. Then the focus is some point p 0; we take p 0. Since OF p, from 11.18 we nd that the directrix is the line x p e (see gure 11.14).

Chapter 11

Conics and Polar Coordinates

162

Figure 11.13

Figure 11.14 y

L m V L X n F n em X x y

F p 0

Now, for a point X (11.19)

y on the curve, we have XL x p e and XF

p 2 y 2

and so equation 11.18 in coordinates is given by (11.20) Case e (11.21)

p 2 y 2

e x p e

ex p

1. Squaring both sides we get x2 2px p2 y2 x2 2px p2 simplifying to y2 4px

This of course is the standard form of a parabola. It also locates the focus and the directrix of a parabola. Proposition 11.2 The focus of the parabola y2 ax is a 4 units on one side of the vertex of the parabola along the axis, and and the directrix intersects the axis a 4 units on the other side. Example 11.5 Find the vertex, focus and directrix of the parabola given by the equation 2x2 6x y 4 0 First we put the equation in standard form. Completing the square, we have (11.22) 2 x2 3x 9 4

9 2

y4

or

3 2

1 1 y 2 2

Thus the vertex is at 3 2 1 2, the axis of the parabola is the line x 3 2 and we have 4p 1 2, so p 1 8. Thus the focus is at 3 2 1 2 1 8 3 2 5 8 and the directrix is the line y 3 8.

Example 11.6 Find the equation of the parabola whose vertex is at 4 2 and whose directrix is the line x 1. Find the focus of this parabola. Since the directrix is a vertical line, the axis is horizontal, so the equation has the form (11.23)
y

2 2

4px 4

11.2

Eccentricity and Foci

163 2 1

since the vertex is at 4 2. Now p is the distance between the vertex and the directrix, so p 5. Thus the equation of the parabola is (11.24)
y

2 2

20x 4

The focus is 5 units to the right of the vertex, so is at 9 2. Example 11.7 Find the equation of the parabola whose focus is the origin and whose vertex is at the point a 0 with a 0. The parabola has its axis the x-axis, and since the vertex is to the right of the focus, the parabola opens to the left. Thus the equation has the form (11.25) y2

4px a 4ax a
e2 x2 2px p2

where p is the distance between focus and vertex. But that is a, so the equation is (11.26) Case e (11.27) which simplies to (11.28)
1

y2

1.Squaring both sides of 11.20 gives us x2 2px p2 y2

e2x2 y2 2p1 ex

Thus, if e 1, this is an ellipse, and if e 1 this is a hyperbola. Notice, because of symmetry in the minor axis, ellipses and hyperbolas have two foci; one on each side of the minor axis. We now want show how to locate the foci of an ellipse given in standard form. Thus we start by putting 11.28 in standard form, and then compare it to the formula of proposition 11.1c. Dividing equation 11.28 by the coefcient of x2 gives us (11.29) x2 2p y2 x 1e 1 e2 0

Now completing the square, we come to (11.30) Comparing this to (11.31)

x

1e

y2 1 e2

p2 1 e 2

x 0 2 y y 0 2
a2 b2

1 p 1 e b 2 ea
1

we see that the center of the ellipse is at p 1 e 0 and a distance of the center from the focus. Then (11.32) c p p 1e e p 1e

e2a2. Let c be the

Chapter 11

Conics and Polar Coordinates a2 b2 . Summarizing

164

and c2

e2 a 2

Proposition 11.3 If an ellipse is in standard form (11.31), with a the major axis, c units away from the center where (11.33) c2 a2 b2

b, then the foci of the ellipse are on

The eccentricity of the ellipse is given by the equations (11.34) b2

1

e2a2

or

ea

The same arguments for the case e

1, the hyperbola, lead to

Proposition 11.4 If a hyperbola is in standard form (11.35)

x

x 0 2 y y 0 2
a2 b2 c2 a2 b2

then the foci of the hyperbola are on the major axis, c units away from the center where (11.36)

The eccentricity of the hyperbola is given by the equations (11.37) b2

e
2

1 a 2

or

ea

Example 11.8 Find the foci of the conic given by the equation x2 4y2 2x First, we complete the square to get the equation in standard form: (11.38)
x

1 2
32

y2 3 2 2

This conic is an ellipse centered at (1,0), with major axis the x-axis, and a2 9 b2 9 4. Thus 2 2 b2 c a 93 4, so c 3 3. This is the distance of the foci from the center (along the major 2 axis), so the foci are at 1 3 2 3 0. Example 11.9 Find the foci of the conic given by the equation y2 x2 4x Complete the squares, and get the standard form (11.39) y2 32 13

x 22 3

1 2. We have c2 a2 b2 2. Thus the foci are at 2

This is a hyperbola with center at 2 0, and major axis the line x c 3 2 is the distance of the foci from the center along the line x The vertices are at 2 3.

18, so 2.

Example 11.10 Find the equation of the ellipse centered at the origin, with a focus at 2 0 and a vertex at 3 0.

11.3

String and Optical Properties of the Conics

165

The equation of an ellipse centered at the origin is (11.40) We are given a (11.41) 3 c 2. Thus b2 x2 a2 a 2 c2 x2 9

y2 b2

5, and the equation is

y2 5

11.3. String and Optical Properties of the Conics

We have seen that the parabola can be dened as the locus of points X equidistant from a given point F and a given line L. The ellipse and the hyperbola have similar denitions. Proposition 11.5 Given two points F1 and F2 and a number a greater than the distance between F1 and F2 , the locus of points X such that (11.42) XF1

XF2

2a

is an ellipse with foci at F1 and F2 and major axis of length 2a. Given an ellipse in standard form, we can verify 11.42 by a lengthy algebraic computation. To show that 11.42 leads to the equation of an ellipse is another algebraic computation beginning this way. Choose coordinates so that the points F1 and F2 lie on the x-axis, and the origin is midway between the points. Then F1 has coordinates c 0, and F2 has coordinates c 0 for some c a. Let X have the coordinates x y. Then 11.42 becomes (11.43)

x c2 y2

c 2 y 2

2a

Eliminate the radicals to verify that we end up with a quadratic equation which is that of an ellipse. We have a similar description of the hyperbola: Proposition 11.6 Given two points F1 and F2 and a positive number a, the locus of points X such that (11.44) is a hyperbola with foci at F1 and F2 . Actually, this is just the branch of the hyperbola which wraps around the focus F2 ; the other branch is given by the equation (11.45) XF2 XF1

XF2

2a

XF1

2a

The optical properties of the conics follow from these string characterizations. Lets start with the parabola. Suppose that the parabola is coated with a light-reecting material. The rays of a beam of light originating far away along the axis of the parabola will approach the parabola along lines parallel to

Chapter 11

Conics and Polar Coordinates

166

Figure 11.15 T ds
x

y X

dx

dy L

LF

its axis. According to the physics of the situation, the angle of reection off the parabola is equal to the angle of incidence. The optical property of the parabola is that these reected rays all meet at the focus. Proposition 11.7 Let X be a point on the parabola, and T the tangent line to the parabola at X. Let LF be the line from the focus to X, and L the line through X parallel to the axis of the parabola. Then the angle between T and LF is equal to the angle between T and L. What we want to show, referring to gure 11.15, is that . From the gure we see that this amounts to showing that . Let us think of the parabola as being traced out by a particle moving to the right at constant velocity 1. This expresses the coordinates x y of the point X as functions of arc length s. The string property of the parabola tells us that (11.46)

c 2 y 2 c 2 y 2

xc

Differentiating with respect to arc length gives us (11.47) which simplies to (11.48)

2x c dx 2y dy ds ds 2 xc c 2 y 2

dx ds

y dx dy 2 y2 ds ds x c

dx ds

Now we do a little trigonometry: (11.49) cos

xc c 2 y 2

sin

y c 2 y 2

cos

dx ds

sin

dy ds

11.3

String and Optical Properties of the Conics

167

Figure 11.16

dx ds dy

1
F1 c 0

1
F2 c 0

so (10) becomes (11.50) cos cos sin sin cos or cos cos , from which we conclude as desired. The optical property of the ellipse is that a ray of light emanating from one focus reects off the ellipse so as to pass through the other focus. Proposition 11.8 Let X be a point on the ellipse, and T the tangent line to the ellipse at X. Let L1 be the line from the focus F1 to X, and L2 the line from the other focus F2 to X. Then the angle between T and L1 is equal to the angle between T and L2 . What we want to show, referring to gure 11.16, is that 2 property, written in the coordinates as shown in the gure: (11.51)

x c2 y2

1 . We start with the string

c 2 y 2

2a

We now differentiate with respect to arc length, and arrive at (11.52)

xc
x c2 y2

y xc y dx dy dx dy 2 y2 ds 2 y2 ds 2 y2 ds ds x c x c x c

We now make the substitutions with the trigonometric functions, but here we have to be careful: in our picture dy and x c are negative, so since the sine and cosine are ratios of lengths, we have (11.53) cos 1

xc

c 2 y 2

sin

dy

ds

dy ds
0

Thus our equation becomes (11.54) or (11.55)

cos 2 cos

cos 2 cos sin 2 cos cos 1 cos sin 1 sin

sin 2 cos cos 1 cos sin 1 sin

Chapter 11

Conics and Polar Coordinates

168

which is cos2 cos1 0, so 2 1 as desired The optical property of the hyperbola is that a ray of light emanating from one focus reects off the opposite branch of the hyperbola so as to appear to have come from the other focus. Proposition 11.9 Let X be a point on the hyperbola, and T the tangent line to the ellipse at X. Let L1 be the line from the focus F1 to X, and L2 the line from the other focus F2 to X. Then the exterior angles between T and L1 and between T and L2 are equal
Figure 11.17: T

L1 F1

L2 F2

11.4. Polar Coordinates

Often a problem can be seen as that of understanding the motion of a particle in the plane relative to a xed point. In such a situation it is desirable to be able to describe a position in terms of the length and the r direction of the line between the two points. These are the polar coor dinates of the point. We consider the xed point as the origin of these x coordinates, and take the positive x-axis as the zero direction. Then any other direction is described by the angle between it and the positive x axis, which we denote as . The distance of a point on this line from the origin is denoted r. These equations relate the cartesian coordinates x y with Figure 11.18 the polar coordinates r : y x2 y2 y r sin r arctan (11.56) x r cos x See gure 11.18 Polar coordinates have two pecularities which we need to get used to. Every value of r determines a point in the plane. However, if r 0, the point is the origin, and doesnt make sense. Secondly, the values r and r 2 , and in fact, r 2n for any n give the same point. This ambiguity is sometimes of value: for example, when discussing the motion of a particle, n tells us how many times the particle has wound around the origin in the counterclockwise sense. Finally, it is also of convenience to let r take negative values, meaning a distance of r in the opposite direction of the ray . Thus r and r determine the same point. We now consider the graphs of equations in polar coordinates.
y

Example 11.11 The equation r a, for a 0 is satised by all points of distance a from the origin, so is polar equation of the circle of radius a centered at the origin.

11.4
Figure 11.19

Polar Coordinates

169

Figure 11.20

Example 11.12 The equation

0 is the line which makes an angle of 0 with the x-axis.

Example 11.13 r a describes the motion of a point which rotates around the origin at angular velocity 1 while moving out along the ray at velocity a. This is the Archimedean spiral and is shown in gure 11.19. Example 11.14 r ea is another spiral, however, the point moves out along the ray at a rate exponential in the rate of rotation. This is the logarithmic spiral and is shown in gure 11.20. Example 11.15 The equation r a cos is the circle of diameter a with center on the x-axis which goes through the origin. For, if we multiply by r we get r2 ar cos , which can now be written in cartesian coordinates (using (11.56)) as (11.57) x2 y2 ax or x a 2
2

a2 4

Given an equation of the form r r , we can often trace out the graph by just studying the behavior of the function r . Lets redo example 11.15 this way. We have this table
0 32 2 2 r 1 0 1 0 1

Figure 11.21

Figure 11.22

X x y

r F

V d

Chapter 11

Conics and Polar Coordinates

170

It will be useful for you to follow the following discussion along the curve in gure 11.21. Between 0 and 2 the point is in the rst quadrant, and as the angle increases it moves toward the origin, reaching there at 2. Then for between 2 and , the point is in the fourth quadrant (because r 0, steadily moving away from the origin until we reach the point weve started with. This looks like a circle, and the argument above (in example 11.15) shows that it is. Note that as moves from to 2 the circle is retraced. Example 11.16 Similarly, the equation r a cos 0 is the circle through the origin of radius a with center on the ray of angle 0 . This amounts to the assertion that any equation of the form (11.58) r a cos b sin

is a circle with the origin the endpoint of one of its diameters (see practice problem 12.1). Example 11.17 If we are given the equation of a curve in cartesian coordinates, we can nd its equation in polar coordinates through the substitution x r cos y r sin . For example (11.59) Equation of a line: r c a cos b sin

For, the general equation of a line is ax by (11.60)

c. After substitution this becomes c

ar cos br cos

which becomes the above when we solve for r. Example 11.18 The polar equation of a conic of eccentricity e, focus at the origin and directrix the line x d is (11.61) Equation of a Conic: r ed 1 e cos e XL , referring to gure 11.22. In polar

To show 11.61, we start with the dening relation XF coordinates this gives us (11.62) r ed x

ed r cos

Solving for r brings us to (11.61). If the gure is rotated by 0 , we just replace with 0 Example 11.19 r a cos 2 . We rst construct the table:

0 4 r a 0

3 4

0 a

Follow this discussion along the graph shown in gure 11.23. This time the curve starts (at 0) at r a, and decreases to zero by 4. Between 4 and 2 r is negative, so the curve is in the third quadrant, and as rotates counterclockwise, r moves away from the origin nally to r a for 2. As increases from 2 the point continues to move toward the origin (in the fourth quadrant), arriving there at 3 2. Moving on, r becomes positive, so we enter the second quadrant with the distance from the origin steadily increasing until, at we are at r a. Since cos is an even function, as we move from to 2 (or what is the same, from to 0), we just get the same curve, reected in the x-axis. This is the four-petalled rose shown in gure 11.24.

11.4
Figure 11.23

Polar Coordinates

171

Figure 11.24

Example 11.20 r a cos 3 is a three-petalled rose. Construct the table of important values between 0 and and argue as in example 11.19. The table is

0 6 r a 0

2 3

5 6

0 a

a
Figure 11.26

That completes the rose; as we proceed from to 2 we traverse the rose again. See gure 11.25.

Figure 11.25

We conclude Proposition 11.10 The graph of the equation r a cosn or r even, and an n petalled rose if n is odd (traversed twice). Limacons. These are the curves dened by the equation r First, we consider the case:a b. We have the table a sinn is a 2n-petalled rose if n is

a b cos .

pi 0 32 2 2 r ab a ab a ab

leading to the graph shown in gure 11.26. As b gets closer and closer to a, the value of r for graph shown in gure 11.27, called the cardioid.

goes to zero. Thus when a

b, we get the

Chapter 11

Conics and Polar Coordinates

172

Figure 11.27

Figure 11.28

Then as b goes beyond a, r becomes negative as gets near , and there is an inner loop of the limacon Example 11.21 r 2 4 cos . Our table is this:
0 pi 32 2 2 r 6 2 2 2 6

When cos 1 2, that is, for 2 3, the value of r is zero, and between these two values r is negative. We get the graph shown in gure 11.28. We have drawn the curve so that it is tangent to 3 for those values of . This is correct, as we will show in the next section. Finally, it is important to note that if the function cos is replaced by cos the curve is reected in the y-axis, and if it is replaced by sin , it is rotated by a right angle.

11.5. Calculus in polar coordinates

Arc length Consider the curve given in polar coordinates by the equation r r . We can calculate the differential ds of arc length by the differential triangle in polar coordinates using this diagram.
Figure 11.29 dr ds rdt

dt

r t

The length of the arc of the circle of radius r subtended by the angle d is rd . The differential triangle is thus a right triangle with side lengths dr and rd . By the pythagorean theorem (11.63) ds2 dr2 r2 d 2

11.5

Calculus in polar coordinates

173

Example 11.22 Find the length of the curve r 2 from 0 to 2 . This curve is a spiral whose distance from the origin increases as the square of the angle. We have dr 2 d , so (11.64) and thus the length is
2 2

ds2

dr2 r2 d 2

4 2 d 2 4 d 2

2 4 2d 2

(11.65)
0

ds
0

4 2d

3 2 2 1 42 3 0

3 1 4 4 2 3

43 2

Area To caculate the area enclosed by a curve given, in polar coordinates, by r differential of area, using gure 11.30.
rd d

r , we calculate the

r r

The area of the wedge given by the increment d is 1 2r2 d . To see this, we start with the area of the circle of radius r : A r2 . Now an angle subtends a segment of the circle which is the ( 2 )th part of the full circle, thus the area of that segment is 1 2r2 . Thus, for d , we get
r (11.66)

dA

1 2 r d 2 31

Figure 11.30

Example 11.23 Find the area enclosed by the cardioid r sin . The area is (11.67) Area 1 2
2 0

31 sin 2 d

9 2

2 0

2 1 2 sin sin d

Now, we know that the integral of sin over an entire period is zero, so we can neglect the middle term. We now use the double angle formula for the last term, and drop the integral of cos2 for the same reason: (11.68) Area 9 2
2

1
0

1 cos 2 d 2

9 2

2 0

3 d 2

27 2

Example 11.24 Find the area inside one petal of the rose r sin 3 . At 0 we have r 0, but then as the angle rotates, r increases to its maximum at 3 2, and then decreases back to zero for 3 . Thus one petal is spanned as ranges from 0 to 3. We now calculate; (11.69) Area 1 2
3
0

sin2 3 d

1 2

3
0

1 cos 6 d 2

1 2

6 cos12

3
0

12

Chapter 11

Conics and Polar Coordinates

174

Figure 11.31 r r

d tan line

Tangents Given the polar equation r r of a curve, we can nd the tangent at any point as follows. First of all, the cartesian coordinates are given by x r cos y r sin . If m is the slope of the tangent line, we have, by the chain rule (11.70) m dy dx dy d dx d

r sin cos ddr

0, then

dr r cos sin d

Notice that, as r 0, the right hand side approaches tan . Thus, if 0 is a value for which r the curve approaches the origin along the ray 0 . Example 11.25 What is the slope of the tangent to the inner loop of the limacon (11.71) at the origin? We nd the values of for which r (11.72) so that 0: 0 or cos r 2 5 cos

2 5 cos

2 5

0 63 radians or 113 6.

Dandelin spheres
In geometry, the Dandelin spheres are one or two spheres that are tangent both to a plane and to a cone that intersects the plane. The intersection of the cone and the plane is a conic section, and the point at which either sphere touches the plane is a focus of the conic section, so the Dandelin spheres are also sometimes called focal spheres.

The Dandelin spheres were discovered in 1822. They are named in honor of the Belgian mathematician Germinal Pierre Dandelin, though Adolphe Quetelet is sometimes given partial credit as well. The Dandelin spheres can be used to prove at least two important theorems. Both of those theorems were known for centuries before Dandelin, but he made it easier to prove them. The first theorem is that a closed conic section (i.e. an ellipse) is the locus of points such that the sum of the distances to two fixed points (the foci) is constant. This was known to Ancient Greek mathematicians such as Apollonius of Perga, but the Dandelin spheres facilitate the proof. The second theorem is that for any conic section, the distance from a fixed point (the focus) is proportional to the distance from a fixed line (the directrix), the constant of proportionality being called the eccentricity. Again, this theorem was known to the Ancient Greeks, such as Pappus of Alexandria, but the Dandelin spheres facilitate the proof. A conic section has one Dandelin sphere for each focus. In particular, an ellipse has two Dandelin spheres, both touching the same nappe of the cone. A hyperbola has two Dandelin spheres, touching opposite nappes of the cone. A parabola has just one Dandelin sphere.

Proof that the curve has constant sum of distances to foci

Consider the illustration, depicting a plane intersecting a cone to form an ellipse (the interior of the ellipse is colored light blue). The two Dandelin spheres are shown, one (G1) above the ellipse, and one (G2) below. The intersection of each sphere with the cone is a circle (colored white).

Each sphere touches the plane at a point, and let us call those two points F1 and F2. Let P be a typical point on the ellipse. Prove: The sum of distances d(F1, P) + d(F2, P) remain constant as the point P moves along the curve. o A line passing through P and the vertex S of the cone intersects the two circles at points P1 and P2. o As P moves along the ellipse, P1 and P2 move along the two circles. o The distance from F1 to P is the same as the distance from P1 to P, because lines PF1 and PP1 are both tangent to the same sphere (G1). o Likewise, the distance from F2 to P is the same as the distance from P2 to P, because lines PF2 and PP2 are both tangent to the same sphere (G2). o Consequently, the sum of distances d(F1, P) + d(F2, P) must be constant as P moves along the curve because the sum of distances d(P1, P) + d(P2, P) also remains constant. This follows from the fact that P lies on the straight line from P1 to P2, and the distance from P1 to P2 remains constant.

This proves a result that had been proved in a different manner by Apollonius of Perga. If (as is often done) one takes the definition of the ellipse to be the locus of points P such that d(F1, P) + d(F2, P) = a constant, then the argument above proves that the intersection of a plane with a cone is indeed an ellipse. That the intersection of the plane with the cone is symmetric

about the perpendicular bisector of the line through F1 and F2 may be counterintuitive, but this argument makes it clear. Adaptations of this argument work for hyperbolas and parabolas as intersections of a plane with a cone. Another adaptation works for an ellipse realized as the intersection of a plane with a right circular cylinder.

Proof of the focus-directrix property

The directrix of a conic section can be found using Dandelin's construction. Each Dandelin sphere intersects the cone at a circle; let both of these circles define their own planes. The intersections of these two planes with the conic section's plane will be two parallel lines; these lines are the directrices of the conic section. However, a parabola has only one Dandelin sphere, and thus has only one directrix. Using the Dandelin spheres, it can be proved that any conic section is the locus of points for which the distance from a point (focus) is proportional to the distance from the directrix. Ancient Greek mathematicians such as Pappus of Alexandria were aware of this property, but the Dandelin spheres facilitate the proof. Neither Dandelin nor Quetelet used the Dandelin spheres to prove the focus-directrix property. The first to do so was apparently Pierce Morton in 1829. The focus-directrix property is essential to proving that astronomical objects move along conic sections around the Sun.

Dandelin

Dandelin Spheres

A floating ball head wearing a dunce-cap/mosquito-net. Where the ocean surface plane meets the mosquito net is the conic section (here it's an ellipse) Where the ball head touches the water is a focus Where the fish kisses the air is a focus. The ball head's hat brim is a directrix plane as is the fish's belt plane. Where the directrix planes meet the ocean surface are two lines called the directrix lines.

file:///Z|/D/misc/mathema/Dandelin.html (1 of 10)5/8/2011 7:13:44 PM

Dandelin

Much of Dandelin's demonstrations rests on the fact that two of a sphere's tangent line segments sharing a far end point are equal. AO = r = OC OB = OB Angle OAB and angle OCB are 90 degrees because AB and CB are tangent line segments. AB = sqrt(OB^2 - r^2) = CB because of Pythagarus' theorem.

file:///Z|/D/misc/mathema/Dandelin.html (2 of 10)5/8/2011 7:13:44 PM

Dandelin

Spokes from the fish's Spokes from the hat belt are all equal length band are all equal for the same reason. because they are tangent lines sharing a far point.

Take away hat spokes from belt spokes and you're left with lamp shade spokes which are all equal.

file:///Z|/D/misc/mathema/Dandelin.html (3 of 10)5/8/2011 7:13:44 PM

Dandelin

Green lines from hat band to ellipse and from focus to the same point are tangent line segments sharing a far point. Same for the orange lines. Green plus orange line segments make a lamp shade spoke.

Tracing the ellipse we see Distance to 1st focus + Distance to 2nd focus = lamp shade spoke.

file:///Z|/D/misc/mathema/Dandelin.html (4 of 10)5/8/2011 7:13:44 PM

Dandelin

Drop a hinge line straight off hat brim to a point on ellipse. Hinge line is a leg of both blue and red right triangles. It forms a constant angle with the cone wall (on which the blue hypotenuse rests) and a constant angle with the cutting plane (on which the red hypotenuse rests). Blue hypotenuse is the point's distance to focus (can you see why?) Red hypotenuse is the point's distance to directrix line. Since the triangles remain the same shape and share a leg the ratio of blue hypotenuse to the red hypotenuse remains constant. This constant is called the ellipse's eccentricity. An ellipse's eccentricity is between 0 and 1. A circle's eccentricity is called zero although it's directrix line is not defined.
file:///Z|/D/misc/mathema/Dandelin.html (5 of 10)5/8/2011 7:13:44 PM

Dandelin

Some mischieveous lout has glued the ballhead brothers together! As before, where the mosquito net meets the ocean is the conic section (This time it's a hyperbola). Where the ballheads touch the ocean are the foci, the hatbrims are the directrix planes and the water line on the hat brims are the directrix lines.

file:///Z|/D/misc/mathema/Dandelin.html (6 of 10)5/8/2011 7:13:44 PM

Dandelin

As before all the little ball's hatspokes are equal. Ditto for the big ball's hat spokes. The big and and little hat spokes together make equal hourglass spokes.

Let a blue line show a point's distance to the near focus and a red line it's distance to the far focus. Both the red and blue line have twins that start from the hat brims. From the hatbrim lines you can see that Distance to 1st focus Distance to 2nd focus = hourglass spoke.

file:///Z|/D/misc/mathema/Dandelin.html (7 of 10)5/8/2011 7:13:44 PM

Dandelin

As with the ellipse, extend a perpendicular hinge from the hat brim to a point on the hyperbola. The hinge is a shared leg of 2 right triangles. The blue hypotenuse is distance to focus and the red hypotenuse is distance to directrix line. As with the ellipse, the ratio of the blue to the red hypotenuse is constant and this constant is the hyperbola's eccentricity. A Hyperbola's eccentricity is always greater than 1.

file:///Z|/D/misc/mathema/Dandelin.html (8 of 10)5/8/2011 7:13:44 PM

Dandelin

A pair of Dandelin spheres define two tangent cones: the thin cone with its apex on the far side of the little sphere, and the fat cone with its apex between the spheres. They also define two sets of cutting planes which are tangent planes to either the thin or fat cone. Planes tangent to the fat cone will cut the thin cone forming ellipses of a specific eccentricity. Planes tangent to the thin cone will cut the fat cone forming hyperbolas of a specific eccentricity. Ellipses' eccentricity = 1/hyperbolas' eccentricity.

file:///Z|/D/misc/mathema/Dandelin.html (9 of 10)5/8/2011 7:13:44 PM

Dandelin

Tilting so the top edge of the mosquito net is parrallel with the ocean, the ball head looks up and down to see no brother or fish is around. This conic section has only one focus. It is the parabola.

Dropping the hinge from the hat brim we can see the blue hypotenuse is the same as the red hypotenuse. Distance to directrix = Distance to focus and the parabola's eccentricity is exactly 1.

file:///Z|/D/misc/mathema/Dandelin.html (10 of 10)5/8/2011 7:13:44 PM