Additional Solutions of the Limiting Case "CLP" of the Problem of Apollonius via Vector Rotations using Geometric Algebra

Additional Solutions of the Limiting Case
“CLP” of the Problem of Apollonius
via Vector Rotations using Geometric Algebra
1

Geometric-Algebra Formulas
for Plane (2D) Geometry
The Geometric Product, and Relations Derived from It
For any two vectors a and b,
a · b = b · a
b ∧ a = −a ∧ b
ab = a · b + a ∧ b
ba = b · a + b ∧ a = a · b − a ∧ b
ab + ba = 2a · b
ab − ba = 2a ∧ b
ab = 2a · b + ba
ab = 2a ∧ b − ba
Deﬁnitions of Inner and Outer Products (Macdonald A. 2010 p. 101.)
The inner product
The inner product of a j-vector A and a k-vector B is
A · B = AB k−j. Note that if j>k, then the inner product doesn’t exist.
However, in such a case B · A = BA j−k does exist.
The outer product
The outer product of a j-vector A and a k-vector B is
A ∧ B = AB k+j.
Relations Involving the Outer Product and the Unit Bivector, i.
For any two vectors a and b,
ia = −ai
a ∧ b = [(ai) · b] i = − [a · (bi)] i = −b ∧ a
Equality of Multivectors
For any two multivectors M and N,
M = N if and only if for all k, M k = N k.
Formulas Derived from Projections of Vectors
and Equality of Multivectors
Any two vectors a and b can be written in the form of “Fourier expansions”
with respect to a third vector, v:
a = (a · ˆv) ˆv + [a · (ˆvi)] ˆvi and b = (b · ˆv) ˆv + [b · (ˆvi)] ˆvi.
Using these expansions,
ab = {(a · ˆv) ˆv + [a · (ˆvi)] ˆvi} {(b · ˆv) ˆv + [b · (ˆvi)] ˆvi}
Equating the scalar parts of both sides of that equation,
2

a · b = [a · ˆv] [b · ˆv] + [a · (ˆvi)] [b · (ˆvi)], and
a ∧ b = {[a · ˆv] [b · (ˆvi)] − [a · (ˆvi)] [b · ˆv]} i.
Also, a2
= [a · ˆv]
2
+ [a · (ˆvi)]
2
, and b2
= [b · ˆv]
2
+ [b · (ˆvi)]
2
.
Reﬂections of Vectors, Geometric Products, and Rotation operators
For any vector a, the product ˆvaˆv is the reﬂection of a with respect to the
direction ˆv.
For any two vectors a and b, ˆvabˆv = ba, and vabv = v2
ba.
Therefore, ˆveθiˆv = e−θi
, and veθi
v = v2
e−θi
.
For any two vectors a and b, ˆvabˆv = ba, and vabv = v2
ba. Therefore,
ˆveθiˆv = e−θi
, and veθi
v = v2
e−θi
.
A useful relationship that is valid only in plane geometry: abc = cba.
Here is a brief proof:
abc = {a · b + a ∧ b} c
= {a · b + [(ai) · b] i} c
= (a · b) c + [(ai) · b] ic
= c (a · b) − c [(ai) · b] i
= c (a · b) + c [a · (bi)] i
= c (b · a) + c [(bi) · a] i
= c {b · a + [(bi) · a] i}
= c {b · a + b ∧ a}
= cba.
3

Additional Solutions of the Special Case “CLP”
of the Problem of Apollonius via Vector
Rotations using Geometric Algebra
Jim Smith
QueLaMateNoTeMate.webs.com
email: nitac14b@yahoo.com
August 27, 2016
Contents
1 Introduction 6
2 A brief review of reflections and rotations in 2D GA 7
2.1 Review of reflections . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1 Reflections of a single vector . . . . . . . . . . . . . . . . 7
2.1.2 Reflections of a bivector, and of a geometric product of
two vectors . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Review of rotations . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3 The Problem of Apollonius, and Its CLP Limiting Case 10
3.1 Observations, and Potentially Useful Elements of the Problem . . 10
3.1.1 Formulating a Strategy . . . . . . . . . . . . . . . . . . . 13
4 Solving the Problem 14
4.1 Solution via Rotations Alone . . . . . . . . . . . . . . . . . . . . 14
4.1.1 Solving for the Solution Circles that Do Not Enclose the
Given Circle . . . . . . . . . . . . . . . . . . . . . . . . . 14
4

4.1.2 Solving for the Solution Circles that Do Enclose the Given
Circle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2 Solution via a Combination of Reﬂections and Rotations . . . . . 24
5 Literature Cited 28
6 Appendix: The Rotations-Only Method Used in the Original
Version of this Document 29
6.1 Identifying the Solution Circles that Don’t Enclose C . . . . . . 29
6.1.1 Formulating a Strategy . . . . . . . . . . . . . . . . . . . 29
6.1.2 Transforming and Solving the Equations that Resulted
from Our Observations and Strategizing . . . . . . . . . . 30
6.2 Identifying the Solution Circles that Enclose C . . . . . . . . . . 36
6.3 The Complete Solution, and Comments . . . . . . . . . . . . . . 38
5

ABSTRACT
This document adds to the collection of solved problems presented in [1]-[6].
The solutions presented herein are not as efficient as those in [6], but they give
additional insight into ways in which GA can be used to solve this problem.
After reviewing, briefly, how reflections and rotations can be expressed and ma-
nipulated via GA, it solves the CLP limiting case of the Problem of Apollonius
in three ways, some of which identify the the solution circles’ points of tangency
with the given circle, and others of which identify the solution circles’ points of
tangency with the given line. For comparison, the solutions that were developed
in [1] are presented in an Appendix.
1 Introduction
This document has been prepared for two very different audiences: for my fellow
students of GA, and for experts who are preparing materials for us, and need
to know which GA concepts we understand and apply readily, and which ones
we do not. In some sense, the solutions presented herein are “leftovers” from
work that was done to find more-efficient solutions, so the document is still in
somewhat of a rough condition, with a certain amount of repetition.
Readers are encouraged to study the following documents, GeoGebra work-
sheets, and videos before beginning:
“Rotations of Vectors via Geometric Algebra: Explanation, and Usage in
Solving Classic Geometric “Construction” Problems”
https://drive.google.com/file/d/0B2C4TqxB32RRdE5KejhQTzMtN3M/view?usp=sharing
“Answering Two Common Objections to Geometric Algebra”
As GeoGebra worksheet
As YouTube video.
“Geometric Algebra: Find unknown vector from two dot products”
As GeoGebra worksheet
As YouTube video
For an more-complete treatment of rotations in plane geometry, be sure to
read Hestenes D. 1999, pp. 78-92. His section on circles (pp. 87-89) is especially
relevant to the present document. Macdonald A. 2010 is invaluable in many
respects, and González Calvet R. 2001, Treatise of Plane Geometry through
Geometric Algebra is a must-read.
The author may be contacted at QueLaMateNoTeMate.webs.com.
6

2 A brief review of reflections and rotations in
2D GA
2.1 Review of reflections
2.1.1 Reflections of a single vector
For any two vectors û and v, the product ûvû is
ûvû = {2û ∧ v + vû} û (2.1)
= v + 2 [(ûi) · v] iû (2.2)
= v − 2 [v · (ûi)] ûi, (2.3)
which evaluates to the reflection of the reflection of v with respect to û (Fig.
2.1).
Figure 2.1: Geometric interpretation of ûvû, showing why it evaluates to the
reflection of v with respect to û.
We also note that because u = u û,
uvu = u2
(ûvû) = u2
v − 2 [v · (ui)] ui. (2.4)
7

2.1.2 Reflections of a bivector, and of a geometric product of two
vectors
The product ûvwû is
ûvwû = û (v · w + v ∧ w) û
= û (v · w) û + û (v ∧ w) û
= û2
(v · w) + û [(vi) · w] iû
= v · w + û [−v · (wi)] (−ûi)
= v · w + û2
[(wi) · v] i
= w · v + w ∧ v
= wv.
In other words, the reflection of the geometric product vw is wv, and does
not depend on the direction of the vector with respect to which it is reflected.
We saw that the scalar part of vw was unaffected by the reflection, but the
bivector part was reversed.
Further to that point, the reflection of geometric product of v and w is
equal to the geometric product of the two vectors’ reflections:
ûvwû = ûv (ûû) wû
= (ûvû) (ûwû) .
That observation provides a geometric interpretation (Fig. 2.2) of why reflecting
a bivector changes its sign: the direction of the turn from v to w reverses.
Figure 2.2: Geometric interpretation of ûvwû, showing why it evaluates to
the reflection of v with respect to û. Note that ûvwû = ûv (ûû) wû =
(ûvû) (ûwû).
8

2.2 Review of rotations
One of the most-important rotations—for our purposes—is the one that is pro-
duced when a vector is multiplied by the unit bivector, i, for the plane: vi
is v’s 90-degree counter-clockwise rotation, while iv is v’s 90-degree clockwise
rotation.
Every geometric product bc is a rotation operator, whether we use it as
such or not:
bc = b c eθi
.
where θ is the angle of rotation from b to c. From that equation, we obtain the
identity
eθi
=
bc
b c
=
b
ˆb
c
ˆc
= ˆbˆc .
Note that a [ûˆv] evaluates to the rotation of a by the angle from û to ˆv, while
[ûˆv] a evaluates to the rotation of a by the angle from ˆv to û.
A useful corollary of the foregoing is that any product of an odd number
of vectors evaluates to a vector, while the product of an odd number of vectors
evaluates to the sum of a scalar and a bivector.
An interesting example of a rotation is the product ûˆvabˆvû. Writing that
product as (ûˆva) (bˆvû), and recalling that any geometric product of two vectors
acts as a rotation operator in 2D (2.2), we can see why ûˆvabˆvû = ab (Fig.
2.3).
Figure 2.3: Left-multiplying a by the geometric product ûˆv rotates a by an
angle equal (in sign as well as magnitude) to that from ˆv to û. Right-multiplying
b by the geometric product ˆvû rotates b by that same angle. The orientations
of a and b with respect to each other are the same after the rotation, and the
magnitudes of a and b are unaffected. Therefore, the geometric products ab
and (ûˆva) (bˆvû) are equal.
We can also write ûˆvabˆvû as ûˆv [ab] ˆvû, giving it the form of a rotation of
the geometric product ab. Considered in this way, the result ûˆvabˆvû = ab is
a special case of an important theorem: rotations preserve geometric products
[?].
9

3 The Problem of Apollonius, and Its CLP Lim-
iting Case
The famous “Problem of Apollonius”, in plane geometry, is to construct all of
circles that are tangent, simultaneously, to three given circles. In one variant
of that problem, one of the circles has inﬁnite radius (i.e., it’s a line). The
Wikipedia article that’s current as of this writing has an extensive description
of the problem’s history, and of methods that have been used to solve it. As
described in that article, one of the methods reduces the “two circles and a line”
variant to the so-called “Circle-Line-Point” (CLP) limiting case:
Given a circle C, a line L, and a point P, construct the circles that
are tangent to C and L, and pass through P.
Figure 3.1: The CLP Limiting Case of the Problem of Apollonius: Given a
circle C, a line L, and a point P, construct the circles that are tangent to C and
L, and pass through P.
3.1 Observations, and Potentially Useful Elements of the
Problem
From the ﬁgure presented in the statement of the problem, we can see that
there are two types of solutions. That is, two types of circles that satisfy the
stated conditions:
• Circles that enclose C;
• Circles that do not enclose C.
As we will see later, there are actually two circles that do enclose the
given circle, and two that do not. We’ll begin by discussing circles that do not
10

enclose it. Most of our observations about that type will also apply, with little
modiﬁcation, to circles that do enclose it.
Based upon our experience in solving other “construction problems” in-
volving tangency, a reasonable choice of elements for capturing the geometric
content of the problem is as shown below:
• Use the center point of the given circle as the origin;
Figure 3.2: Initial step in choosing elements for capturing the geometric content
of the problem. (See text for explanation.)
• Capture the perpendicular distance from c1’s center to the given line in
the vector h;
• Express the direction of the given line as ±ˆhi.
• Label the solution circle’s radius and its points of tangency with C and L
as shown in Fig. 3.3:
In deriving our solution, we’ll
use the same symbol —for
example, t —to denote both a
point and the vector to that
point from the origin. We’ll rely
upon context to tell the reader
whether the symbol is being
used to refer to the point, or to
the vector.
Now, we’ll express key features of the problem in terms of the elements
that we’ve chosen. First, we know that we can write the vector s as s =
h+λˆhi, where λ is some scalar. We can even identify that scalar, speciﬁcally, as
ˆt · ˆhi . We also know that the points of tangency t and s are equidistant (by
r2) from the center point of the solution circle. Combining those observations,
we can equate three expressions for the vector s:
s = (r1 + r2)ˆt + r2
ˆh = h + λˆhi = h + ˆt · ˆhi ˆhi. (3.1)
Examining that equation, we observe that we can obtain an expression for
11

Figure 3.3: Further steps in choosing elements for capturing the geometric
content of the problem. (See text for explanation.)
r2 in terms of known quantities by “dotting” both sides with ˆh:
(r1 + r2)ˆt + r2
ˆh · ˆh = h + λˆhi · ˆh
(r1 + r2)ˆt · ˆh + r2
ˆh · ˆh = h · ˆh + λ ˆhi · ˆh
(r1 + r2)ˆt · ˆh + r2 = h + 0;
∴ r2 =
h − r1
ˆt · ˆh
1 + ˆt · ˆh
. (3.2)
The denominator of the expression on the right-hand side might catch our
attention now because one of our two expressions for the vector s, namely
s = (r1 + r2)ˆt + r2
ˆh
can be rewritten as
s = r1
ˆt + r2
ˆt + ˆh .
That fact becomes useful (at least potentially) when we recognize that ˆt + ˆh
2
=
2 1 + ˆt · ˆh . Therefore, if we wish, we can rewrite Eq. (3.2) as
r2 = 2



h − r1
ˆt · ˆh
ˆt + ˆh
2


 .
Those results indicate that we should be alert to opportunities to simplify
expressions via appropriate substitutions invoving ˆt + ˆh and 1 + ˆh · ˆt
As a ﬁnal observation, we note that when a circle is tangent to other objects,
there will be many angles that are equal to each other. For example, the angles
whose measures are given as θ in Figs. 3.4 and 3.5:
12

Figure 3.4: A pair of equal angles that might provide a basis for solving the
problem via rotations.)
We also see in Fig. 3.5 that points T and S are reflections of each other
with respect to a line that passes through the center of the solution circle, and
has the direction of (t − s) i. References [1]-[6] showed how GA enables us to
make use of the reflections and rotations that we’ve noted.
Many of the ideas that we’ll employ here will also be used when we treat
solution circles that do enclose C.
3.1.1 Formulating a Strategy
Now, let’s combine our observations about the problem in a way that might
lead us to a solution. Our previous experiences in solving problems via vector
rotations suggest that we should equate two expressions for the rotation eθi
:
t − p
t − p
s − p
s − p
=
t − s
t − s
−ˆhi
=
s − t
s − t
ˆhi . (3.3)
We’ve seen elsewhere that we will probably want to transform that equation
into one in which some product of vectors involving our unknowns t and s is
equal either to a pure scalar, or a pure bivector. By doing so, we may find some
way of identifying either t or s.
We’ll keep in mind that although Eq. (3.3) has two unknowns (the vectors
t and s), our expression for r2 (Eq. (3.2)) enables us to write the vector s in
terms of the vector ˆt.
Therefore, our strategy is to
13

Figure 3.5: Another pair of equal angles that might provide a basis for solving
the problem via rotations. Note, too, that the points T and s are reflections of
each other with respect to the vector (t − s) i.
• Equate two expressions, in terms of the unknown vectors t and s, for the
rotation eθi
;
• Transform that equation into one in which on side is either a pure scalar
or a pure bivector;
• Watch for opportunities to simplify equations by substituting for r2; and
• Solve for our unknowns.
4 Solving the Problem
Our first solution, of which two variants are presented, uses rotations alone.
The second uses a combination of reflections and rotations. Reference [1] can
be studied for comparison.
4.1 Solution via Rotations Alone
4.1.1 Solving for the Solution Circles that Do Not Enclose the Given
Circle
We’ll start by examining Fig. 4.1, which is based upon Fig. 3.5. Using ideas
presented in Section 3.1, we can write
s − p
s − p
t − p
t − p
= eφi
= ˆh


i ˆt + ˆh
ˆt + ˆh

 ,
14

Figure 4.1: Elements used in finding the solution circles that do not enclose the
given one. The vector s has been written as (r1 + r2)ˆt + r2
ˆh, and we’ve made
use of the fact that i ˆt + ˆh has the same direction as i (s − t). (Compare Fig.
3.5.)
from which
s − p
s − p
t − p
t − p
= ˆh


i ˆt + ˆh
ˆt + ˆh

 ,
ˆh
s − p
s − p
t − p
t − p
=


i ˆt + ˆh
ˆt + ˆh

 ,
ˆh
s − p
s − p
t − p
t − p
=


i ˆt + ˆh
ˆt + ˆh

 ,
ˆh
s − p
s − p
t − p
t − p
ˆt + ˆh = i ˆt + ˆh , and
ˆh [s − p] [t − p] ˆt + ˆh = i ˆt + ˆh s − p t − p ,
the right-hand side of which is a bivector. Therefore,
ˆh [s − p] [t − p] ˆt + ˆh 0 = 0.
We can save ourselves some work by expanding the left-hand side in a way that
let’s us take advantage of what we’ve learned about the form in which GA can
express reflections:
ˆh [s − p] [t − p]ˆt 0 + ˆh [s − p] [t − p] ˆh 0 = 0. (4.1)
Note that the product ˆh [s − p] [t − p] ˆh is just the reflection of [s − p] [t − p]
with respect to ˆh, and is therefore equal to [t − p] [s − p]. Let’s find the scalar
part of that product, then return to the first term in 4.1. We’ll use the substitu-
tions t = r1
ˆt; r2 =
h − r1
ˆh · ˆt
1 + ˆh · ˆt
; r1 +r2 =
r1 + h
1 + ˆh · ˆt
; and s = (r1 + r2)ˆt+r2
ˆh
15

to write
[t − p] [s − p] = r1
ˆt − p
r1 + h
1 + ˆh · ˆt
ˆt +
h − r1
ˆh · ˆt
1 + ˆh · ˆt
ˆh − p
= r1
ˆt − p


(r1 + h )ˆt + h − r1
ˆh · ˆt ˆh − 1 + ˆh · ˆt p
1 + ˆh · ˆt

 .
That last step—putting everything in the last factor over the common
denominator 1 + ˆh · ˆt —is quite a natural one to take, but turns out to be
unnecessary. By taking it, I complicated solutions in other problems needlessly,
by forming awkward products like p · ˆt ˆh · ˆt . Later in this solution we’ll see
why that step is unnecessary, but for now, let’s go back to
[t − p] [s − p] = r1
ˆt − p
r1 + h
1 + ˆh · ˆt
ˆt +
h − r1
ˆh · ˆt
1 + ˆh · ˆt
ˆh − p ,
then continue to expand the right-hand side:
[t − p] [s − p] = r1
ˆt − p
r1 + h
1 + ˆh · ˆt
ˆt +
h − r1
ˆh · ˆt
1 + ˆh · ˆt
ˆh − p
=
r1 (r1 + h )
1 + ˆh · ˆt
ˆtˆt +


r1 h − r1
ˆh · ˆt
1 + ˆh · ˆt

ˆtˆh
−r1
ˆtp −
r1 + h
1 + ˆh · ˆt
pˆt −
h − r1
ˆh · ˆt
1 + ˆh · ˆt
pˆh + p2
,
the scalar part of which is
[t − p] [s − p] 0 =
r1 (r1 + h )
1 + ˆh · ˆt
+


r1 h − r1
ˆh · ˆt
1 + ˆh · ˆt

 ˆh · ˆt
−r1p · ˆt −
r1 + h
1 + ˆh · ˆt
p · ˆt −
h − r1
ˆh · ˆt
1 + ˆh · ˆt
p · ˆh + p2
.
16

Now, let’s ﬁnd ˆh [s − p] [t − p]ˆt 0 from 4.1:
ˆh [s − p] [t − p]ˆt = ˆh
r1 + h
1 + ˆh · ˆt
ˆt +
h − r1
ˆh · ˆt
1 + ˆh · ˆt
ˆh − p r1
ˆt − p ˆt
=
r1 (r1 + h )
1 + ˆh · ˆt
ˆhˆtˆtˆt −
r1 + h
1 + ˆh · ˆt
ˆhˆtpˆt
+


r1 h − r1
ˆh · ˆt
1 + ˆh · ˆt

 ˆhˆhˆtˆt −


r1 h − r1
ˆh · ˆt
1 + ˆh · ˆt

 ˆhˆhpˆt
−r1
ˆhpˆtˆt + ˆhppˆt
=
r1 (r1 + h )
1 + ˆh · ˆt
ˆhˆt −
r1 + h
1 + ˆh · ˆt
ˆhˆtpˆt
+


r1 h − r1
ˆh · ˆt
1 + ˆh · ˆt

 −


r1 h − r1
ˆh · ˆt
1 + ˆh · ˆt

 pˆt
−r1
ˆhp + p2 ˆhˆt.
The scalar part of ˆhˆtpˆt is ˆhˆt 2p · ˆt − ˆtp 0 = 2 p · ˆt ˆh · p − ˆh · p.
Therefore,
ˆh [s − p] [t − p]ˆt 0 =
r1 (r1 + h )
1 + ˆh · ˆt
ˆh · ˆt
−
r1 + h
1 + ˆh · ˆt
2 p · ˆt ˆh · p − ˆh · p
+


r1 h − r1
ˆh · ˆt
1 + ˆh · ˆt

 −


r1 h − r1
ˆh · ˆt
1 + ˆh · ˆt

 p · ˆt
−r1
ˆh · p + p2 ˆh · ˆt.
As we noted in 6.1,
ˆh [s − p] [t − p]ˆt 0 + ˆh [s − p] [t − p] ˆh 0 = 0.
Adding our the expressions that we’ve just developed for ˆh [s − p] [t − p]ˆt 0
17

and ˆh [s − p] [t − p] ˆh 0, we obtain
r1 (r1 + h )
1 + ˆh · ˆt
+


r1 h − r1
ˆh · ˆt
1 + ˆh · ˆt

 ˆh · ˆt
−r1p · ˆt −
r1 + h
1 + ˆh · ˆt
p · ˆt −
h − r1
ˆh · ˆt
1 + ˆh · ˆt
p · ˆh + p2
+
r1 (r1 + h )
1 + ˆh · ˆt
ˆh · ˆt −
r1 + h
1 + ˆh · ˆt
2 p · ˆt ˆh · p − ˆh · p
+


r1 h − r1
ˆh · ˆt
1 + ˆh · ˆt

 −


r1 h − r1
ˆh · ˆt
1 + ˆh · ˆt

 p · ˆt
−r1
ˆh · p + p2 ˆh · ˆt = 0. (4.2)
Note the pairs of expressions in red and blue: those pairs sum, respectively, to
r1 (r1 + h ) and r1 h − r1
ˆh · ˆt . That fact is one of the reasons why we
didn’t need to put all of the terms over the common denominator 1+ ˆh·ˆt. (See
the discussion that follows 4.1.)
After simpliﬁcation, 4.2 reduces to
2 (r1 + h ) p · ˆt − p2
− r1
2 ˆh · ˆt = 2 h r1 + r1
2
+ p2
.
We saw equations like this last one many times in Smith J A 2016. There,
we learned to solve those equations by grouping the dot products that involve
t into a dot product of t with a linear combination of known vectors:
2 (r1 + h ) p − p2
− r1
2 ˆh
A linear combination of ˆh and p
·ˆt = 2 h r1 + r1
2
+ p2
. (4.3)
The geometric interpretation of Eq. (4.3) is that 2 h r1 + r1
2
+ p2
is the
projection of the vector 2 (r1 + h ) p − p2
− r1
2 ˆh upon ˆt. Because we want
to ﬁnd t, and know 2 (r1 + h ) p − p2
− r1
2 ˆh, we’ll transform Eq. (4.3)
into a version that tells us the projection of the vector t upon 2 (r1 + h ) p −
p2
− r1
2 ˆh.
First, just for convenience, we’ll multiply both sides of Eq. (4.3) by r1 h :
2 r1 h + h2
p − p2
− r1
2
h · t = 2h2
r1
2
+ r1 h r1
2
+ p2
.
Next, we’ll use the symbol “w” for the vector 2 r1 h + h2
p − p2
− r1
2
h ,
and write
w · t = 2h2
r1
2
+ r1 h r1
2
+ p2
.
18

Finally, because P w (t), the projection of the vector t upon w is (t · ˆw) ˆw,
we have
P w (t) =
2h2
r1
2
+ r1 h r1
2
+ p2
w
ˆw. (4.4)
As we learned in Smith J A 2016, Eq. (4.4) tells us that Eq (4.3) has two
solutions. That is, there are two circles that are tangent to L and pass through
the point P, and are also tangent to C without enclosing it:
Figure 4.2: The two solution circles that do not enclose the given circle, C.
Having identified P w (t), the points of tangency with C and L can be
determined using methods shown in Smith J A 2016, as can the equations for
the corresponding solution circles.
To round off our treatment of solution circles that don’t enclose C, we
should note that we derived our solution starting from equations that express
the relationship between C, L, P, and the smaller of the two solution circles.
You may have noticed that the larger solution circle does not bear quite the
same relationship to L, P, and C as the smaller one. To understand in what
way those relationships differ, please see the discussion in 6.
4.1.2 Solving for the Solution Circles that Do Enclose the Given
Circle
To introduce some new ideas and show how to employ useful GA identities,
we’ll follow a somewhat different route in this section than we did when finding
the solution circles that don’t enclose the given one. Instead of expressing s
—the point of tangency with the given line—as a linear combination of ˆt and
19

ˆh, we’ll express it as a linear combination of h and ˆhi: s = h + λˆh, where λ
is a scalar. Because λˆhi is the projection of (r1 − r3)ˆt upon ˆhi (Fig. 4.3), we
can identify λ, specifically, as (r1 − r3) ˆt · ˆhi .
Figure 4.3: The two solution circles that do enclose the given circle, C. The vec-
tor ˆt − ˆh i has the same direction as (t − s) i. Because λˆhi is the projection
of (r1 − r3)ˆt upon ˆhi, we can identify λ, specifically, as (r1 − r3) ˆt · ˆhi .
At this point, we can foresee needing to derive an expression for r3 that’s
analogous to that which we presented in Eq. 3.2 for r2. We’ll do so by equating
two expressions for s, then ”dotting” with ˆh:
(r1 − r3)ˆt + r3
ˆh = h + λˆh
(r1 − r3)ˆt + r3
ˆh · ˆh = h + λˆhi · ˆh
(r1 − r3)ˆt · ˆh + r3 = h
∴ r3 =
h − r1
1 − ˆh · ˆt
, and r1 − r3 =
r1 − h
1 − ˆh · ˆt
. (4.5)
Now we’re ready to begin to solve for t. We could start by equating
two expressions for eωi, but just to be different we’ll note that the product
t − p
t − p
s − p
s − p
, which is equal to eωi, is a rotation operator that brings
ˆt − ˆh i into alignment with ˆh:


ˆt − ˆh i
ˆt − ˆh

 t − p
t − p
s − p
s − p
= ˆh,
20

which we then transform via


ˆt − ˆh i
ˆt − ˆh

 t − p
t − p
s − p
s − p
= ˆh,
i
ˆt − ˆh
ˆt − ˆh
t − p
t − p
s − p
s − p
= −ˆh,
ˆt − ˆh
ˆt − ˆh
t − p
t − p
s − p
s − p
= iˆh,
ˆt − ˆh
ˆt − ˆh
t − p
t − p
s − p
s − p
ˆh = i,
ˆt − ˆh [t − p] [s − p] ˆh 0 = 0. (4.6)
In addition, we’ll use the Fourier expansion ˆt = ˆt · ˆh ˆh + ˆt · ˆhi ˆhi
to transform the factor ˆt − ˆh:
ˆt − ˆh = ˆt · ˆh ˆh + ˆt · ˆhi ˆhi
= ˆh · ˆt − 1 ˆh + ˆhi · ˆt ˆhi.
Now, we’ll substitute that expression in (4.6), and make an initial expan-
sion:
ˆt − ˆh [t − p] [s − p] ˆh 0 = 0,
ˆh · ˆt − 1 ˆh + ˆhi · ˆt ˆhi [t − p] [s − p] ˆh 0 = 0,
from which
ˆh · ˆt − 1 ˆh [t − p] [s − p] ˆh 0
+ ˆhi · ˆt ˆhi [t − p] [s − p] ˆh 0 = 0
and
ˆh · ˆt − 1 ˆh [t − p] [s − p] ˆh 0
− ˆhi · ˆt ˆh [t − p] [s − p] ˆh i 0 = 0.
Both of the terms within the angle brackets are reﬂections of the product
[t − p] [s − p] with respect to ˆh. Therefore,
ˆh · ˆt − 1 [s − p] [t − p] 0
− ˆhi · ˆt {[s − p] [t − p]} i 0 = 0,
from which
ˆh · ˆt − 1 [s − p] · [t − p]
− ˆhi · ˆt {[s − p] ∧ [t − p]} i = 0. (4.7)
21

We’ll treat each of those terms separately. From (4.5), and from s =
h + λˆhi = h + (r1 − r3) ˆt · ˆhi ˆhi, ˆh · ˆt − 1 [s − p] · [t − p] is
ˆh · ˆt − 1 h +
r1 − h
1 − ˆh · ˆt
ˆt · ˆhi ˆhi − p · r1
ˆt · ˆh ˆh + r1
ˆt · ˆhi ˆhi − p .
We needn’t put the terms in
curly brackets over the common
denominator 1 − ˆh · ˆt, because
that denominator is canceled by
the factor ˆh · ˆt − 1, leaving -1.
After expansion, that result simpliﬁes to
ˆh · ˆt − 1 [s − p] · [t − p] = − h ˆh · p ˆh · ˆt − h ˆhi · p ˆhi · ˆt
+r1 h ˆh · ˆt
2
+ r1 h ˆhi · ˆt
2
+2r1
ˆhi · p ˆhi · ˆt − r1 h ˆh · ˆt
+ h ˆh · p − r1
2 ˆhi · ˆt
2
− r1
ˆh · p ˆh · ˆt
2
+r1
ˆh · p ˆh · ˆt − r1
ˆhi · p ˆhi · ˆt ˆh · ˆt
+p2 ˆh · ˆt − p2
.
We could simplify even further by recognizing that
ˆh · p ˆh · ˆt + ˆhi · p ˆhi · ˆt = p · ˆt,
so that the terms in red reduce to − h p · ˆt. However, we will not do so
because our goal is to obtain a solution in the form of a linear combination of
ˆh and ˆhi. The terms in blue are a diﬀerent story: they sum to r1 h because
ˆh · ˆt
2
+ ˆhi · ˆt
2
= ˆt
2
= 1. Therefore,
ˆh · ˆt − 1 [s − p] · [t − p] = − h ˆh · p ˆh · ˆt − h ˆhi · p ˆhi · ˆt
+r1 h + 2r1
+ h ˆh · p − r1
2 ˆhi · ˆt
2
− r1
ˆh · p ˆh · ˆt
2
+r1
+p2 ˆh · ˆt − p2
. (4.8)
Now, we’ll treat the term ˆhi · ˆt {[s − p] ∧ [t − p]} i from Eq. (4.7). I’ll
go into some detail so that you may see the interesting uses of GA identities
that arise.
Using the same substitutions for r3 and λ as we did when deriving (4.8),
ˆhi · ˆt {[s − p] ∧ [t − p]} i is
ˆhi · ˆt h +
r1 − h
1 − ˆh · ˆt
ˆt · ˆhi ˆhi − p ∧ r1
ˆt − p .
22

Expanding that product using the identity a ∧ b ≡ [(ai) · b] i ≡ − [a · (bi)] i,
we ﬁnd that
ˆhi · ˆt {[s − p] ∧ [t − p]} i = ˆhi · ˆt r1
ˆh ˆhi · ˆt i − ˆh ˆhi · p i
−
r1 (r1 − h ) ˆhi · ˆt ˆh · ˆt i
1 − ˆh · ˆt
−
(r1 − h ) ˆhi i · p ˆhi · ˆt i
1 − ˆh · ˆt
− r1
ˆh · ˆt (pi) · ˆh i
−r1
ˆhi · ˆt (pi) · ˆhi i i. (4.9)
As was the case when we expanded and simpliﬁed ˆh · ˆt − 1 [s − p]·[t − p],
the presence of 1 − ˆh · ˆt in the denominators of some of the terms is not
a problem because the only terms with that denominator contain the factor
ˆhi · ˆt in their numerators. Thus, when we multiply through by the factor
ˆhi · ˆt that’s outside the curly brackets, the numerators of those terms will
contain the factor ˆhi · ˆt
2
, which is equal to 1 − ˆh · ˆt
2
, and therefore to
1 − ˆh · ˆt 1 + ˆh · ˆt . The factor 1 − ˆh · ˆt in that product will cancel with
the 1 − ˆh · ˆt in the denominator, leaving 1 + ˆh · ˆt in the numerator.
Making use of that observation about the cancellation of the denominators
in Eq. (4.9), we can further expand and simplify terms in that equation to
obtain
ˆhi · ˆt {[s − p] ∧ [t − p]} = − r1 h ˆhi · ˆt
2
− h ˆhi · p ˆhi · ˆt
− r1 (r1 − h ) ˆh · ˆt − r1 (r1 − h ) ˆh · ˆt
2
+ (r1 − h ) ˆh · p + (r1 − h ) ˆh · p ˆh · ˆt
− r1 (pi) · ˆh ˆhi · ˆt ˆh · ˆt
−r1
ˆh · p ˆhi · ˆt
2
. (4.10)
In (Eq. (4.7)), we noted that
ˆh · ˆt − 1 [s − p] · [t − p] − ˆhi · ˆt {[s − p] ∧ [t − p]} i = 0.
Substituting in that equation the expressions for ˆh · ˆt − 1 [s − p]·[t − p] and
23

ˆhi · ˆt {[s − p] ∧ [t − p]} i in Eqs. (4.8) and (4.10),
h ˆh · p ˆh · ˆt − h ˆhi · p ˆhi · ˆt
+r1 h + 2r1
+ h ˆh · p − r1
2 ˆhi · ˆt
2
− r1
ˆh · p ˆh · ˆt
2
+r1
+p2 ˆh · ˆt − p2
+r1 h ˆhi · ˆt
2
− h ˆhi · p ˆhi · ˆt
−r1 (r1 − h ) ˆh · ˆt − r1 (r1 − h ) ˆh · ˆt
2
+ (r1 − h ) ˆh · p + (r1 − h ) ˆh · p ˆh · ˆt
−r1 (pi) · ˆh ˆhi · ˆt ˆh · ˆt − r1
ˆh · p ˆhi · ˆt
2
= 0.
In simplifying that equation, we again make use of the identity ˆh · ˆt
2
+
ˆhi · ˆt
2
= ˆt
2
= 1. The result is
2r1 h − h ˆh · p ˆh · ˆt − h ˆhi · p ˆhi · ˆt − r1
2
+ (2r1 − h ) ˆh · p ˆh · ˆt + (2r1 − h ) ˆhi · p ˆhi · ˆt
+p2 ˆh · ˆt − p2
− r1
2 ˆh · ˆt = 0.
As we did for the solution circles that don’t enclose the given, we transform the
equation that we’ve just obtained into one involving the dot product of t with
a linear combination of known vectors:
2 ( h − r1) ˆh · p + r1
2
− p2 ˆh
+ 2 ( h − r1) ˆhi · p ˆhi · t = 2r1 h − p2
− r1
2
. (4.11)
4.2 Solution via a Combination of Reﬂections and Rota-
tions
For more information on the ideas used in this solution, please see [2]. The key
ideas that we’ll use, with reference to Fig. 4.2, are:
• The vector s can be expressed as the sum of h and a scalar multiple (λ )
of ˆhi;
• Our goal will be to determine the values of λ for the four solution circles;
• The points s and t are reﬂections of each other with respect to the vector
i (s − t), which is the mediatrix of the base of the isosceles triangle tC2s;
24

• The isosceles triangle C1C2M is similar to tC2s. Therefore, the
points s and t are reflections of each other with respect to the vector
i h + r1
ˆh + λˆhi . The advantage of using that vector (as opposed to
s − t) for expressing reflections is that it is written in terms of λ, and has
the origin (C1) as its starting point.
• Similarly, the vectors ˆt and −ˆh are reflections of each other with respect
to i h + r1
ˆh + λˆhi .
Figure 4.4: Elements used in using a combination of reflections and rotations
to identify the two solution circles that do not enclose the given circle, C.
We’ll incorporate those observations in our usual technique of equating two
expressions for the same rotation, which in this case will be eθi:
t − p
t − p
s − p
s − p
=


i h + r1
ˆh + λˆhi
h + r1
ˆh + λˆhi

 ˆh ;
∴ [t − p] [s − p] ˆh h + r1
ˆh + λˆhi = a bivector, and
ts − tp − ps + p2 ˆh h + r1
ˆh + λˆhi 0 = 0.
We’ll expand that result, initially, as
(ts) ˆh h + r1
ˆh + λˆhi 0
− (tp) ˆh h + r1
ˆh + λˆhi 0
− (ps) ˆh h + r1
ˆh + λˆhi 0
+ p2 ˆh h + r1
ˆh + λˆhi 0 = 0. (4.12)
25

To expand each of those four terms, we´ll write s as s = h+λˆhi = h ˆh+λˆhi,
and h + r1
ˆh + λˆhi as (r1 + h ) ˆh + λˆhi. In addition, we’ll write ˆt as the
reﬂection of −ˆh with respect to i h + r1
ˆh + λˆhi :
ˆt =


i h + r1
ˆh + λˆhi
h ˆh + r1
ˆh + λˆhi

 −ˆh


i h + r1
ˆh + λˆhi
h ˆh + r1
ˆh + λˆhi


=
(r1 + h ) ˆh + λˆhi ˆh (r1 + h ) ˆh + λˆhi
(r1 + h )
2
+ λ2
,
from which
t =
r1 (r1 + h ) ˆh + λˆhi ˆh (r1 + h ) ˆh + λˆhi
(r1 + h )
2
+ λ2
. (4.13)
The factors in red constitute a
rotation operator. See 2.3.
Now, for the expansions of each of the four terms in (4.12).
(ts) ˆh h + r1
ˆh + λˆhi
=



(r1 + h )
2
+ λ2



h ˆh + λˆhi ˆh (r1 + h ) ˆh + λˆhi
=



r1 (r1 + h ) ˆh + λˆhi ˆh (r1 + h ) ˆh + λˆhi h ˆh + λˆhi ˆh (r1 + h ) ˆh + λˆhi
(r1 + h )
2
+ λ2



= r1 (r1 + h ) ˆh + λˆhi h ˆh + λˆhi
∴ (ts) ˆh ˆh h + r1
ˆh + λˆhi 0 = r1
2
h + r1h2
+ r1λ2
. (4.14)
The factors in red constitute a
rotation operator. See 2.3.
(tp) ˆh h + r1
ˆh + λˆhi
=



(r1 + h )
2
+ λ2



[p] ˆh (r1 + h ) ˆh + λˆhi
=



r1 (r1 + h ) ˆh + λˆhi ˆh (r1 + h ) ˆh + λˆhi [p] ˆh (r1 + h ) ˆh + λˆhi
(r1 + h )
2
+ λ2



= r1 (r1 + h ) ˆh + λˆhi [p]
∴ (tp) ˆh h + r1
ˆh + λˆhi 0 = r1 (r1 + h ) ˆh · p − λr1
ˆhi · p . (4.15)
(ps) ˆh h + r1
ˆh + λˆhi
= [p] h ˆh + λˆhi ˆh (r1 + h ) ˆh + λˆhi
∴ (ps) ˆh h + r1
ˆh + λˆhi 0 = r1 h + h2
p · ˆh − λ (r1 + 2 h ) p · ˆh + λ2
p · ˆh.
(4.16)
26

p2 ˆh h + r1
ˆh + λˆhi =
(r1 + h ) p2
+ λp2
i
∴ p2 ˆh h + r1
ˆh + λˆhi 0 = (r1 + h ) p2
. (4.17)
If we wished, we could now
make the substitution
λ = (r1 + r2) ˆt · ˆhi (see
Fig. 3.3), then proceed to
transform this quadratic into
Eq. (4.1.1).
Using the scalar parts that we identified in Eqs. (4.14)-(4.17), we find that
(4.12) reduces to the following quadratic in λ:
r1 + p · ˆh λ2
− 2 (r1 + h ) p · ˆhi λ
+ (r1 + h ) p2
+ r1 h − (r1 + h )
2
p · ˆh = 0.
Its solutions are
λ =
(r1 + h ) p · ˆhi ± (p2 − r1
2) (r1 + h ) [(h − p)] · ˆh
r1 + p · ˆh
. (4.18)
The interpretation of that result is that there are two solution circles that don’t
enclose C, and each of those circles is tangent to L at its own distance λ from
the point at which h intersects L.
To find the solution circles that do enclose C, we start from Fig. 4.5.
Figure 4.5: Elements used in using a combination of reflections and rotations
to identify the solution circles that enclose the given circle, C.
This time, we must express ˆt as a reflection of ˆh with respect to i h − r1
ˆh + λˆhi .
From there, the procedure is the same as we used to find the solution circles
that do not enclose C. But note: ˆt for solution circles that do enclose C is the re-
flection of +ˆh with respect to i h − r1
ˆh + λˆhi , whereas ˆt for solution circles
27

that don’t enclose C is the reflection of −ˆh with respect to i h + r1
ˆh + λˆhi .
The result, for solution circles that do enclose C, is
λ =
(r1 − h ) p · ˆhi ± (p2 − r1
2) ( h − r1) [(h − p)] · ˆh
r1 − p · ˆh
. (4.19)
5 Literature Cited
References
[1] “Solution of the Special Case ‘CLP’ of the Problem of Apollo-
nius via Vector Rotations using Geometric Algebra”. Available at
http://vixra.org/abs/1605.0314.
[2] “The Problem of Apollonius as an Opportunity for Teaching Students to
Use Reflections and Rotations to Solve Geometry Problems via Geometric
(Clifford) Algebra”. Available at http://vixra.org/abs/1605.0233.
[3] J. Smith, “Rotations of Vectors Via Geometric Algebra: Explanation, and
Usage in Solving Classic Geometric ‘Construction’ Problems” (Version of
11 February 2016). Available at http://vixra.org/abs/1605.0232 .
[4] J. Smith, “A Very Brief Introduction to Reflections in 2D Geometric
Algebra, and their Use in Solving ‘Construction’ Problems”. Available
at http://vixra.org/abs/1606.0253.
[5] J. Smith, “Three Solutions of the LLP Limiting Case of the Problem
of Apollonius via Geometric Algebra, Using Reflections and Rotations”.
Available at http://vixra.org/abs/1607.0166.
[6] J. Smith, “Simplified Solutions of the CLP and CCP Limiting Cases of the
Problem of Apollonius via Vector Rotations using Geometric Algebra”.
Available at http://vixra.org/abs/1608.0217.
GeoGebra Worksheets and Related Videos (by title, in alphabetical order):
”Answering Two Common Objections to Geometric Algebra”
GeoGebra worksheet: http://tube.geogebra.org/m/1565271
YouTube video: https://www.youtube.com/watch?v=oB0DZiF86Ns.
”Geometric Algebra: Find unknown vector from two dot products”
GeoGebra worksheet:
http://tube.geogebra.org/material/simple/id/1481375
YouTube video:
https://www.youtube.com/watch?v=2cqDVtHcCoE
Books and articles (according to author, in alphabetical order):
28

González Calvet R. 2001. Treatise of Plane Geometry through Geometric
Algebra. 89.218.153.154:280/CDO/BOOKS/Matem/Calvet.Treatise.pdf.
Retrieved 30 December 2015.
Hestenes D. 1999. New Foundations for Classical Mechanics (Second
Edition). Kluwer Academic Publishers, Dordrecht/Boston/London.
Macdonald A. 2010. Linear and Geometric Algebra. CreateSpace Independent
Publishing Platform, ASIN: B00HTJNRLY.
Smith J. A. 2014. “Resoluciones de ’problemas de construcción’ geométricos
por medio de la geometr´ıa clásica, y el Álgebra Geométrica”,
http://quelamatenotemate.webs.com/3%20soluciones
%20construccion%20geom%20algebra%20Cliffod.xps
2016. “Rotations of Vectors via Geometric Algebra: Explanation, and
Usage in Solving Classic Geometric “Construction” Problems”
https://drive.google.com/file/d/0B2C4TqxB32RRdE5KejhQTzMtN3M/view?usp=sharing
6 Appendix: The Rotations-Only Method Used
in the Original Version of this Document
6.1 Identifying the Solution Circles that Don’t Enclose C
Many of the ideas that we’ll employ here will also be used when we treat solution
circles that do enclose C.
6.1.1 Formulating a Strategy
Now, let’s combine our observations about the problem in a way that might
lead us to a solution. Our previous experiences in solving problems via vector
rotations suggest that we should equate two expressions for the rotation eθi
:
t − p
t − p
s − p
s − p
=
t − s
t − s
−ˆhi
=
s − t
s − t
ˆhi . (6.1)
We’ve seen elsewhere that we will probably want to transform that equation
into one in which some product of vectors involving our unknowns t and s is
equal either to a pure scalar, or a pure bivector. By doing so, we may find some
way of identifying either t or s.
We’ll keep in mind that although Eq. (6.1) has two unknowns (the vectors
t and s), our expression for r2 (Eq. (3.2)) enables us to write the vector s in
29

terms of the vector ˆt.
Therefore, our strategy is to
• Equate two expressions, in terms of the unknown vectors t and s, for the
rotation eθi
;
• Transform that equation into one in which on side is either a pure scalar
or a pure bivector;
• Watch for opportunities to simplify equations by substituting for r2; and
• Solve for our unknowns.
6.1.2 Transforming and Solving the Equations that Resulted from
Our Observations and Strategizing
For convenience, we’ll present our earlier figure again:
By examining that figure, we identified and equated two expressions for
the rotation eθi
, thereby obtaining Eq. (6.1):
t − p
t − p
s − p
s − p
=
s − t
s − t
ˆhi .
We noted that we might wish at some point to make the substitution
s = [r1 + r2]ˆt + r2
ˆh
= r1 +
h − r1
ˆt · ˆh
1 + ˆt · ˆh
ˆt +
h − r1
ˆt · ˆh
1 + ˆt · ˆh
ˆh. (6.2)
We also noted that we’ll want to transform Eq. (6.1) into one in which one
side is either a pure scalar or a pure bivector. We should probably do that
30

transformation before making the substitution for s. One way to effect the
transformation is by left-multiplying both sides of Eq. (6.1) by s − t, then by
ˆh, and then rearranging the result to obtain
ˆh [s − t] [t − p] [s − p] = s − t t − p s − p i (6.3)
This is the equation that we sought to obtain, so that we could now write
ˆh [s − t] [t − p] [s − p] 0 = 0. (6.4)
Next, we need to expand the products on the left-hand side, but we’ll want
to examine the benefits of making a substitution for s first. We still won’t, as
yet, write s in terms of ˆt. In hopes of keeping our equations simple enough
for us to identify useful simplifications easily at this early stage, we’ll make the
substitution
s = (r1 + r2)ˆt + r2
ˆh,
rather than making the additional substitution (Eq. (3.2)) for r2. Now, we can
see that s − t = r2
ˆt + ˆh . Using this result, and t = r1
ˆt, Eq. (6.4) becomes
ˆh r2
ˆt + ˆh r1
ˆt − p (r1 + r2)ˆt + r2
ˆh − p 0 = 0.
Now here is where I caused myself a great deal of unnecessary work in
previous versions of the solution by plunging in and expanding the product
that’s inside the 0 without examining it carefully. Look carefully at the last
factor in that product. Do you see that we can rearrange it to give the following?
ˆh r2
ˆt + ˆh r1
ˆt − p r2
ˆt + ˆh + r1
ˆt − p
After rearrangement
0 = 0.
That result is interesting, but is it truly useful to us? To answer that question,
let’s consider different ways in which we might expand the product, then find
its scalar part.
If we effect the multiplications in order, from left to right, we’re likely to
end up with a confusing mess. However, what if we multiply the last three
factors together? Those three factors, together, compose a product of the form
ab [a + b]:
r2
ˆt + ˆh
a
r1
ˆt − p
b


r2
ˆt + ˆh
a
+ r1
ˆt − p
b


 .
The expansion of ab [a + b] is
ab [a + b] = aba + b2
a
= 2 (a · b) a − a2
b + b2
a (among other possibilites).
That expansion evaluates to a vector, of course. Having obtained the cor-
responding expansion of the product r2
ˆt + ˆh r1
ˆt − p r2
ˆt + ˆh + r1
ˆt − p , we’d
31

then “dot” the result with ˆh to obtain ˆh r2
ˆt + ˆh r1
ˆt − p r2
ˆt + ˆh + r1
ˆt − p 0.
We know, from the solutions to Problem 6 in Smith J A 2016 , that such a ma-
neuver can work out quite favorably. So, let’s try it.
Expanding r2
ˆt + ˆh r1
ˆt − p r2
ˆt + ˆh + r1
ˆt − p according to the identity
ab [a + b] = 2 (a · b) a − a2
b + b2
a, we obtain, initially,
2 r2
2 ˆt + ˆh · r1
ˆt − p ˆt + ˆh − r2
2 ˆt + ˆh
2
r1
ˆt − p + r2 r1
ˆt − p
2
ˆt + ˆh
When we’ve completed our expansion and dotted it with ˆh, we’ll set the result
to zero, so let’s divide out the common factor r2 now:
2 r2
ˆt + ˆh · r1
ˆt − p ˆt + ˆh − r2
ˆt + ˆh
2
r1
ˆt − p + r1
ˆt − p
2
ˆt + ˆh
Recalling that ˆt + ˆh
2
= 2 1 + ˆh · ˆt , the preceding becomes
2 r2
ˆt + ˆh · r1
ˆt − p ˆt + ˆh − 2r2 1 + ˆh · ˆt r1
ˆt − p + r1
ˆt − p
2
ˆt + ˆh .
This is the form that we’ll dot with ˆh. Having done so, the factor ˆt+ ˆh becomes
1 + ˆh · ˆt. Then, as planned, we set the resulting expression equal to zero:
2 r2
ˆt + ˆh · r1
ˆt − p 1 + ˆh · ˆt − 2r2 1 + ˆh · ˆt r1
ˆh · ˆt − ˆh · p + r1
ˆt − p
2
1 + ˆh · ˆt = 0.
Next, we’ll rearrange that equation to take advantage of the relation r2 =
h − r1
ˆh · ˆt
1 + ˆh · ˆt
(see Eq. (3.2)). We’ll show the steps in some detail:
2 r2
ˆt + ˆh · r1
ˆt − p 1 + ˆh · ˆt − 2r2 1 + ˆh · ˆt r1
ˆh · ˆt − ˆh · p + r1
ˆt − p
2
1 + ˆh · ˆt = 0.
2r2 1 + ˆh · ˆt ˆt + ˆh · r1
ˆt − p − r1
ˆh · ˆt + ˆh · p + r1
ˆt − p
2
1 + ˆh · ˆt = 0
2
h − r1
ˆh · ˆt
1 + ˆh · ˆt
1 + ˆh · ˆt ˆt + ˆh · r1
ˆt − p − r1
ˆh · ˆt + ˆh · p + r1
ˆt − p
2
1 + ˆh · ˆt = 0
2 h − r1
ˆh · ˆt ˆt + ˆh · r1
ˆt − p − r1
ˆh · ˆt + ˆh · p + r1
ˆt − p
2
1 + ˆh · ˆt = 0
2 h − r1
ˆh · ˆt r1 − p · ˆt + r1
ˆh · ˆt − ˆh · p − r1
ˆh · ˆt + ˆh · p + r1
ˆt − p
2
1 + ˆh · ˆt = 0
2 h − r1
ˆh · ˆt r1 − p · ˆt + r1
ˆt − p
2
1 + ˆh · ˆt = 0.
Now that the dust has settled from the r2 substitution, we’ll expand r1
ˆt − p
2
,
then simplify further:
2 h − r1
ˆh · ˆt r1 − p · ˆt + r1
2
− 2r1p · ˆt + p2
1 + ˆh · ˆt = 0
−2r1
2
+ 2r1p · ˆt + r1
2
− 2r1
ˆt · p + p2 ˆh · ˆt − 2 h p · ˆt + 2 h r1 + r1
2
− 2r1p · ˆt + p2
= 0
p2
− r1
2 ˆh · ˆt − 2 (r1 + h ) p · ˆt + r1
2
+ p2
+ 2 h r1 = 0.
We saw equations like this last one many times in Smith J A 2016. There,
we learned to solve those equations by grouping the dot products that involve
t into a dot product of t with a linear combination of known vectors:
2 (r1 + h ) p − p2
− r1
2 ˆh
A linear combination of ˆh and p
·ˆt = 2 h r1 + r1
2
+ p2
. (6.5)
32

The geometric interpretation of Eq. (6.5) is that 2 h r1 + r1
2
+ p2
is the
projection of the vector 2 (r1 + h ) p − p2
− r1
2 ˆh upon ˆt. Because we want
to find t, and know 2 (r1 + h ) p − p2
− r1
2 ˆh, we’ll transform Eq. (6.5)
into a version that tells us the projection of the vector t upon 2 (r1 + h ) p −
p2
− r1
2 ˆh.
First, just for convenience, we’ll multiply both sides of Eq. (6.5) by r1 h :
2 r1 h + h2
p − p2
− r1
2
h · t = 2h2
r1
2
+ r1 h r1
2
+ p2
.
Next, we’ll use the symbol “w” for the vector 2 r1 h + h2
p − p2
− r1
2
h ,
and write
w · t = 2h2
r1
2
+ r1 h r1
2
+ p2
.
Finally, because P w (t), the projection of the vector t upon w is (t · ˆw) ˆw,
we have
P w (t) =
2h2
r1
2
+ r1 h r1
2
+ p2
w
ˆw. (6.6)
As we learned in Smith J A 2016, Eq. (6.6) tells us that Eq (6.5) has two
solutions. That is, there are two circles that are tangent to L and pass through
the point P, and are also tangent to C without enclosing it:
Having identified P w (t), the points of tangency with C and L can be
determined using methods shown in Smith J A 2016, as can the equations for
the corresponding solution circles.
To round off our treatment of solution circles that don’t enclose C, we
should note that we derived our solution starting from equations that express
33

the relationship between C, L, P, and the smaller of the two solution circles.
You may have noticed that the larger solution circle does not bear quite the
same relationship to L, P, and C as the smaller one. To understand in what
way those relationships differ, let’s examine the following figure.
By equating two expressions for the rotation eψi
, we’d find that
s − p
s − p
t − p
t − p
= ˆhi
t − s
t − s
.
Compare that result to the corresponding equation for the smaller of the solution
circles:
t − p
t − p
s − p
s − p
=
s − t
s − t
ˆhi .
We followed up on that equation by transforming it into one in which ˆh was at
one end of the product on the left-hand side. The result was Eq. (6.3):
ˆh [s − t] [t − p] [s − p] = s − t t − p s − p i.
We saw the advantages of that arrangement when we proceeded to solve for t.
All we had to do in order to procure that arrangement was to left-multiply both
sides of the equation
t − p
t − p
s − p
s − p
=
s − t
s − t
ˆhi
by s − t, and then by ˆh.
To procure a similar arrangement starting from the equation that we wrote
for the larger circle, using the angle ψ ,
s − p
s − p
t − p
t − p
= ˆhi
t − s
t − s
,
34

we could left-multiply both sides of the equation by ˆh, then right-multiply by
s − t, giving
ˆh [s − p] [t − p] [t − s] = s − p t − p t − s i.
Using the same ideas and transformations as for the smaller circle, we’d then
transform the product [s − p] [t − p] [t − s] into
r2
ˆt + ˆh + r1
ˆt − p r1
ˆt − p r2
ˆt + ˆh .
By comparison, the product that we obtained for the smaller circle was
r2
ˆt + ˆh r1
ˆt − p r2
ˆt + ˆh + r1
ˆt − p .
In both cases, the products that result from the expansion have the forms
ˆt + ˆh r1
ˆt − p ˆt + ˆh and r1
ˆt − p
2
ˆt + ˆh , so the same simplifications
work in both, and give the same results when the result is finally dotted with
ˆh and set to zero.
The above having been said, let’s look at another way of obtaining an equa-
tion, for the larger circle, that has the same form as Eq. (6.3). For convenience,
we’ll present the figure for the larger circle again:
Instead of beginning by equating two expressions for the rotation eψi
, we’ll
equate two expressions for e−ψi
:
t − p
t − p
s − p
s − p
=
t − s
t − s
ˆhi .
After left-multiplying both sides by t−s, then by ˆh, and rearranging, the result
would be identical to Eq. (6.3), except for the algebraic sign of the right-had
35

side, which —because it’s a bivector—would drop out when we took the scalar
part of both sides.
However, the difference in the sign of that bivector captures the geometric
nature of the difference between the relationships of the large and small circles
to L, P, and C. That difference in sign is also reflected in the positions, with
respect to the vector w, of the solution circles’ points of tangency t:
6.2 Identifying the Solution Circles that Enclose C
These solution circles can be found by modifying, slightly, the ideas that we
used for finding solution circles that don’t enclose C. Here, too, we’ll want to
express the radius r3 of the solution circle in terms of the vector t. Examining
the next figure,
36

we see that s = (r1 − r3)ˆt+r3
ˆh, and also that s = h+λˆhi. By equating those
two expressions, dotting both sides with ˆh, and then solving for r3, we ﬁnd that
r3 =
h − r1
ˆt · ˆh
1 − ˆt · ˆh
.
As was the case when we found solution circles that didn’t enclose C, we’ll
want to equate expressions for two rotations that involve the unknown points
of tangency t and s. For example, through the angles labeled φ, below:
t − p
t − p
s − p
s − p
=
t − s
t − s
−ˆhi
=
s − t
s − t
ˆhi ,
Left-multiplying that result by s − t, and then by ˆh,
ˆh [s − t] [t − p] [s − p] = s − t t − p s − p i
∴ ˆh [s − t] [t − p] [s − p] 0 = 0,
which is identical to Eq. (6.4). To solve for t, we use exactly the same technique
that we did when we identiﬁed the solution circles that don’t surround C, with
r3 and 1 − ˆh · ˆt taking the place of r2 and 1 − ˆh · ˆt, respectively. The result is
z · t = 2h2
r1
2
− r1 h r1
2
+ p2
,
where z = 2 h2
− r1 h p − p2
− r1
2
h . Thus,
P z (t) =
2h2
r1
2
− r1 h r1
2
+ p2
z
ˆz. (6.7)
Again, there are two solution circles of this type:
37

6.3 The Complete Solution, and Comments
There are four solution circles: two that enclose C, and two that don’t:
38

Additional Solutions of the Limiting Case "CLP" of the Problem of Apollonius via Vector Rotations using Geometric Algebra

More Related Content

What's hot (18)

Viewers also liked (20)

Similar to Additional Solutions of the Limiting Case "CLP" of the Problem of Apollonius via Vector Rotations using Geometric Algebra (20)

More from James Smith (10)

Recently uploaded (20)

Additional Solutions of the Limiting Case "CLP" of the Problem of Apollonius via Vector Rotations using Geometric Algebra