Complex Numbers in Geometry

Sebastian Jeon
December 3, 2016

1 The Complex Plane

1.1 Definitions
I assume familiarity with most, if not all, of the following definitions. Some knowledge of linear
algebra is also recommended, but not required.
Subsequently, let i be the imaginary unit satisfying i2 = −1. Define the set of complex num-
bers C = {z | z = a + bi, a, b ∈ R} where a is the √ real part of z and b is the imaginary part.
The magnitude of a given z = a + bi ∈ C is |z| = a2 + b2 . The conjugate of z = a + bi will be
z = a−bi, implying the property |z|2 = z ·z. As an exercise, show that z + w = z +w and zw = zw.

Every z ∈ C can be expressed as |z| · (cos θ + i sin θ) for some angle θ ∈ [0, 2π), or alternatively
|z| · eiθ . This angle θ will be referred to as the argument of z.

Lastly, any complex number satisfying z n − 1 = 0 will be referred to as an n th root of unity.

It can easily be verified that these numbers take the form e2kπi/n for 0 ≤ k ≤ n − 1.

1.2 Applications to the Complex Plane

The complex plane assigns a complex number to every point in the plane such that the point P
with Cartesian coordinates (a, b) is assigned a + bi. The counterpart to the xy-axes in the complex
plane are the real and imaginary axes, respectively. From this definition, the similarity between
the complex plane and the Cartesian plane should be evident. Consequently, the representation
z = |z| · (cosθ + i sin θ) = |z|eiθ corresponds to polar coordinates in the Cartesian plane.

Throughout the lecture, the lowercase letter p of a point P will correspond to the complex co-
ordinate, or affix, of P unless otherwise noted.

As an exercise, verify that:

- z is the reflection of z over the real axis
-z=z

1.3 Transformations in the Complex Plane

Geometrically, in order to find the coordinates of the sum of two complex numbers, one may simply
perform a vector head-to-tail addition.

Multiplication of complex numbers is slightly more interesting. For any complex numbers z, w, the
map z → zw corresponds to a spiral similarity (composition of a dilation and rotation) about the
origin. Furthermore, the magnitudes and arguments of zw are determined independently in this
transformation. More specifically, we have
|wz| = |w||z| and arg(w) + arg(z) = arg(wz).
Convince yourself that multiplication by a real number is equivalent to a dilation and multiplica-
tion by eiθ is equivalent to a counterclockwise rotation of θ.
az + b
As a side-note, any transformation z → where a, b, c, d ∈ C, |ac| > 0, ad − bc 6= 0 is a
cz + d
Mobius transformation.

1
2 Metric Propositions
One significant advantage of complex numbers over Cartesian coordinates is that every point is
assigned a single number as opposed to an ordered pair, allowing concise algebraic expressions of
geometric concepts. From these fundamental propositions, the advantages of complex numbers in
various configurations should manifest themselves.

Proposition 2.1. (Angle Between Two Lines).

Let the angle formed by two lines AB, CD in the clockwise direction from AB to CD be θ =
](AB, CD). Then
a−b c−d
= eiθ .
|a − b| |c − d|
Proof. Translate the segments such that B, D coincide with the origin, with A0 = a − b, C 0 = c − d.
a−b
|a−b| r1 eiθ1 /r1
Then c−d = iθ2 /r
= eiθ1 /eiθ2 = ei(θ1 −θ2 ) = eiθ .
|c−d|
r2 e 2

This result serves as the basis for many useful corollaries.

Corollary 2.2. (Alternate form of 1)

Squaring both sides of Proposition 2.1 yields the more applicable form
a−b c−d
= e2iθ .
a−b c−d

Corollary 2.3. (Collinearity)

Points A, B, C are collinear iff
a−b c−b
= .
a−b c−b

Proof. It remains to show that the angle between AB, CB is 0, which follows from letting θ = 0
 
b−a b−a
in Corollary 2.2. Note that this can also be rearranged to = .
b−c b−c
Corollary 2.4. (Equation of a Line)
From Corollary 2.3, letting c be a variable point gives the equation of a line.

Corollary 2.5. (Perpendicularity)

Segments AB, CD are perpendicular iff
a−b c−d
=− .
a−b c−d
Proof. Let θ = π/2 in Corollary 2.2.

Proposition 2.6. (Directly Similar Triangles)

4ABC ∼ 4DEF and ABC, DEF are similarly oriented iff
a−b d−e
= .
a−c d−f
AB DE
Sketch. Consider the magnitudes of both sides of the equation to get = . Now dividing
AC DF
this with the given condition and applications of Proposition 2.1, we get ∠BAC = ∠EDF.

Corollary 2.7 (Spiral Similarity)

ad − bc
The center of spiral similarity taking AB → CD is given by .
a+d−b−c
p−a p−c
Proof. By Proposition 2.5, it suffices to solve = for p.
p−b p−d

2
 Proposition 2.8 (Cyclicity)
Points A, B, C, D are concyclic iff
(a − c)(b − d) (a − d)(b − c)
= .
(a − c)(b − d) (a − d)(b − c)

Sketch. Rewrite ∠ACB = ∠ADB with Proposition 2.1.

Proposition 2.9. The area of triangle ABC is

a a 1
i
b b 1 = a(b − c) − a(b − c) + (bc − bc).
4
c c 1
Sketch. Expand with Shoelace formula and Cartesian coordinates; details are left as an exercise to
the reader.

Proposition 2.10. (Reflection About a Line)

The reflection of P with respect to line AB, denoted by z, satisfies
(a − b)p + ab − ab
z= .
a−b
Sketch. The proof of this uses the fact that linear transformations preserve reflections, so taking
z−a
z → and noting that AB is mapped to the real axis (more specifically, the line segment
b−a
containing 0 and 1) reaches the conclusion.

Proposition 2.11. (Circumcenter Formula)

a aa 1
b bb 1
c cc 1
The circumcenter of 4ABC, denoted by x, satisfies x = .
a a 1
b b 1
c c 1
Sketch. Consider the radius R of the circumcircle, Then |x − a|2 = |x − b|2 = |x − c|2 = r2
so expanding we get a system of equations in x, x, r2 − |x|2 which we can solve with Cramer’s rule.

3 The Unit Circle

One of the fundamental properties of the complex plane that make enormous computations viable
1
is the fact that for any point p on the unit circle, p = . Since any circle can be mapped to the unit
p
circle via a composition of translation and dilation, it is often useful to let a cumbersome circle to
be the unit circle.

Proposition 3.1 (Equation of a Chord)

If AB is a chord of the unit circle, the equation of line AB is given by
z = a + b − abz.
z−a a−b z−a a−b
Proof. From Corollary 2.4, = ⇐⇒ = = −ab and upon expanding we
z−a a−b 1 1
z− −b
a a
get the desired result. Remark on the simplicity of this equation, possible only by this property of
the unit circle.

Corollary 3.2 (Chord Intersection)

Let AB, CD be two chords on the unit circle. If P = AB ∩ CD then
ab(c + d) − cd(a + b)
p=
ab − cd

3
if ab − cd 6= 0.
Sketch. From Proposition 3.1, we know that p is a solution to both
(
p = a + b − abp
p = c + d − cdp.

Subtracting these two equations and solving for p, we consequently get p after some simplification.

Corollary 3.3 (Tangent Intersection)

If the tangents to the unit circle at A, B intersect at P, then
2ab
p= .
a+b
Proof. Note that a tangent is simply a degenerate chord; substituting AA, BB into Corollary 3.2
we get the desired.

Another nice property of unit circles regards fundamental triangle centers of triangles inscribed in
the unit circle.

Proposition 3.4 (Orthocenter, Centroid, Nine-Point Center)

For any triangle ABC inscribed in the unit circle, its orthocenter, centroid, and nine-point center
is given by
a+b+c a+b+c
a + b + c, ,
3 2
respectively.
a+b+c
Proof. Note that from knowledge of Cartesian coordinates, the centroid is regardless of
3
origin. The rest follows from the Euler Line.

Proposition 3.5 (Incenter, Excenters, and Midpoints of Arcs)

Let ABC be a triangle inscribed in the unit circle, and let the complex coordinates of A, B, C
be a2 , b2 , c2 for complex numbers a, b, c. Let A1 , B1 , C1 have coordinates −bc, −ca, −ab, I have
coordinate −ab − bc − ca, and finally Ia , Ib , Ic have coordinates ca + ab − bc, ab + bc − ca, bc + ca − ab
respectively. Then A1 , B1 , C1 are midpoints of arcs BC, CA, AB, I is the incenter of ABC, and
Ia , Ib , Ic are the A, B, C-excenters.

Proof. Since | − bc| = |b| ∗ |c| = 1 it lies on the unit circle. Furthermore, by Proposition 2.1
b2 −a2
2i∠A1 AB b2 −a2−a2 b2 b c
we have e = −bc−a62 2
= = − . Similarly e2i∠A1 AC = − so A1 is on the A-angle
a bc c b
−bc−a2
bisector, implying that it is indeed the midpoint of arc. We obtain similar results for b1 , c1 . To
prove that I has coordinate −ab − bc − ca, note that I is the orthocenter of A1 B1 C1 and apply
Proposition 3.4. The proof for the excenters is left to the reader as an exercise.

Proposition 3.6 (Regular Polygons and Roots of Unity)

Let P1 , P2 , · · · , Pn in the complex plane satisfy Pk = ω k , where ω = e2πi/n . Then P1 P2 · · · Pn is a
regular n-gon.

Proof. Evidently |pk | = 1, so all of the points lie on the unit circle. Furthermore, it can easily be
confirmed that |pk+1 − pk | is constant for 1 ≤ k ≤ n, pn+1 = p1 and similarly that ∠Pk Pk+1 Pk+2
is constant, done.

4
4 Some Instructive Examples
When bashing, it is helpful to let a prominent circle in the diagram be the unit circle, or otherwise
configure the diagram such that the coordinates have nice values. However, it should be noted
that when doing so, the diagram should be general enough to encompass all possible cases.

Problem 4.1 Let ABC be a triangle with circumcenter O. Suppose D and E lie on AB and
AC, respectively, such that O lies on DE. Let M and N be the midpoints of CD and BE, respec-
tively. Prove that ∠M ON = ∠BAC. (Iran 2004 / WOOT)

Proof. WLOG let ω, the circumcircle of ABC, be the unit circle, and further let DE coincide
with the real axis. (Convince yourself that these assumptions are sufficiently general.) Then from
a−b a−d
the definition of point D, A, B, D are collinear =⇒ = . But since D lies on the real
a−b a−d
axis, d = d and solving for d yields
a−b a−d
=
a−b a−d
(a − b)(a − d) = (a − d)(a − b)
aa − ad − ab + bd = aa − ab − ad + bd
d(b − a + a − b) = −ab + ab
−ab + ba −a2 + b2 (b − a)(b + a) b+a
d= = 2 2
= = .
−a + a + b − b −a b + b + ab − a (b − a)(ab + 1) ab + 1
d+c abc + c + b + a abc + a + b + c
Hence, the midpoint of CD is simply m = = , and similarly n =
2 2(ab + 1) 2(ac + 1)
m  n  abc+c+b+a abc+a+b+c
2i]M ON 2(ab+1) 2(ac+1) ac + 1 ab + 1 ac + 1 c + abc c
Thus e = · = abc+c+b+a
· = · = · = .
n m 2(ac+1)
abc+a+b+c ab + 1 ac + 1 ab + 1 b + abc b
2(ab+1)
a−c a−b a2 c − ac2 b−a −ac c
But e2i]BAC = · = · 2 = = , so we may conclude.
a−c a−b c−a a b − ab2 −ab b

Problem 4.2 Let ABC be an acute triangle with AB, AC, BC. Denote by O, H the circum-
center and orthocenter of 4ABC. Suppose that the circumcircle of 4AHC intersects AB again
at M and the circumcircle of 4AHB intersects AC again at n. Prove that the circumcenter of
4M N H lies on line OH. (APMO 2010)

Proof. This problem illustrates the importance of synthetic observations when complex bash-
ing. Once we assume that ω is the unit circle, it is very difficult to obtain coordinates for points
M, N without massive computations. Hence, we need some sort of synthetic insight.

Let D be the foot of the C-altitude, and P be the circumcenter of 4M N H. Easily obtain the
coordinates of D by reflecting C across AB, and taking the midpoint of CC 0 .

Note that ∠CM B = π − ∠AM B = π − ∠AHB = ∠C, so BM C is isosceles, and it thus fol-
a + b + c − abc
lows that D is the midpoint of M C. Now d = =⇒ m = a + c − abc, and similarly
2
we get n = a+b−abc. Note that O = 0, H = a+b+c so a translation of −a−b−c will still yield the
−b(a + c) −c(a + b) 0
circumcenter of M N H on line OH. Then m0 = , and similarly n0 = , h = 0.
c b
Then by the circumcenter formula,
  
b(a+c) c(a+b) a+b a+c


0 c b ac − ab
p =
 c(a+b) 
 b(a+c) 
a+c a+b
ab b − ac c

5
 a+b a+c


0 (a + c)(a + b) ac − ab bc(a + b + c)
p = c b
=− .
b2 + bc + c2


(a + c)(a + b) ab2 − ac2
0
It suffices to show p , o, h collinear, but this follows from

p0 − o − bc(a+b+c)
b2 +bc+c2 abc(a + b + c) h−o
0
= bc+ca+ab
= = ,
p −o − a(b2 +bc+c2 ) ab + bc + ca h−o

and we may conclude.

Problem 4.3 In 4ABC, let O be the circumcenter of (ABC), A1 to be the antipode of A WRT
(ABC), A2 to be the reflection of O across AB. Let OA be the circumcenter of A1 A2 O. Define
OB , OC similarly. Prove that OA , OB , OC are collinear. (Lemmas in Olympiad Geometry)

Proof. Evidently OA ∈ BC and similar, since the perpendicular bisector of OA2 is BC by def-
BOA COB AOC
inition of reflection. Now it suffices to show that · · = 1 by Menelaus. To do
COA AOB BOC
so, we use complex numbers. Let (ABC) be the unit circle so that A1 = −a, A2 = b + c. Then
a1 a2 −a(b + c)(−a − b − c)
OA = = .
a1 a2 − a1 a2 a(b + c) − a(b + c)
a+b+c+ab+ab2 +abc
|BOA | a(b+c)−a(b+c) a + b + c + ab + ab2 + abc a + b + c + ab + ab2 + abc
Now = a+b+c+abc+ac2 +abc
= = ·
|COA | a + b + c + abc + ac2 + abc a + b + c + abc + ac2 + abc
a(b+c)−a(b+c)
|abc| |c| a b + ab + abc + a + b3 + b2 c
2 2 2
|c| |b 
+c| · |a2 + ab + b2 | |c| · |a2 + ab + b2 |
= · 2 2 2 3 2
= · 2 2
= , at
|abc| |b| a c + abc + ac + a b + c + bc |b| |b 
+ c| · |a + ac + c |
 |b| · |a2 + ac + c2 |
which point it is clear that multiplying cyclically will give the desired result.

Problem 4.4 In 4ABC with incenter I, the incircle is tangent to CA, AB at E, F. The reflection of
E, F across I are G, H. Let Q = BC ∩GH, and let M be the midpoint of BC. Prove that IQ ⊥ IM.

Proof. Let the incircle be tangent to BC at D, and let the incircle be the unit circle. This
gives nice coordinates from chord intersection formulas, and moreover we get g = −e and h = −f.
2ef
Then A = EE ∩ F F so a = and similar coordinates can be derived for B and C. Next
e+f

d2 (g + h) − gh(d + d) −d2 (e + f ) − 2def

Q = BC ∩ GH = DD ∩ GH =⇒ = ,
d2 − gh d2 − ef
   
2f d 2de
f +d + d+e f d2 + 2def + ed2
M =⇒ = .
2 (d + f )(d + e)
q m q q
By the perpendicularity condition, it suffices to show =− ⇐⇒ =− .
q m m m
But  
de+df +2ef
q −d 2
d −ef d2 + ed + df + ef
1
+ 1
+ 1
+ 1 q
d2 ed df ef
=   =− = 1 1 =− ,
m df +2ef +de
d (d+f d2 − ef d2 − ef
m
)(d+e)

so we may conclude.

5 Practice Problems
Problem 5.1 A quadrilateral ABCD is inscribed in unit circle.If P = AB∩CD and Q = AD∩BC,
then prove that
p · q + q · p = 2.

Problem 5.2 Let ABC be a triangle with incricle Γ and let D, E, F be the tangency points

6
of Γ with BC, CA, AB, respectively. Let K be the orthocenter of triangle DEF and let ΓA , ΓB , ΓC
be circles centered at A, B, C with radii AD, BE, CF, respectively. Prove that K is the radical
center of ΓA , ΓB , ΓC . (AMSP Geo 3)

Problem 5.3 Oscar is drawing diagrams with trash can lids and sticks. He draws a triangle
ABC and a point D such that DB and DC are tangent to the circumcircle of ABC. Let B 0 be
the reflection of B over AC and C 0 be the reflection of C over AB. If O is the circumcenter of
DB 0 C 0 , help Oscar prove that AO is perpendicular to BC. (ELMO, 2016)

Problem 5.4 Quadrilateral AP BQ is inscribed in circle ω with ∠P = ∠Q = 90◦ and AP =

AQ < BP . Let X be a variable point on segment P Q. Line AX meets ω again at S (other than
A). Point T lies on arc AQB of ω such that XT is perpendicular to AX. Let M denote the mid-
point of chord ST . As X varies on segment P Q, show that M moves along a circle. (USAJMO,
2015)

Problem 5.5 Let P be a point in the plane of triangle ABC, and γ a line passing through
P . Let A0 , B 0 , C 0 be the points where the reflections of lines P A, P B, P C with respect to γ
intersect lines BC, AC, AB, respectively. Prove that A0 , B 0 , C 0 are collinear. (USAMO, 2012)

Problem 5.6 Let ABC be a triangle and let P be a point on its circumcircle. Let X, Y, Z
be the reflections of P over lines BC, CA, AB respectively. Then X, Y, Z, H are collinear where H
is the orthocenter of 4ABC. (Steiner Line)

Problem 5.7 Prove that the incircle of 4ABC is tangent to the nine-point circle of 4ABC.
(Feuerbach)

Problem 5.8 Let ABCD be a cyclic quadrilateral centered at O and let E = AC ∩ BD,
F = AB ∩ CD, and G = AD ∩ BC. Prove that O is the orthocenter of 4EF G. (Brokard)