Beyond complex numbers

Johanna Rämö
July 22, 2010

These notes are written for the Goldsmiths’ Company Mathematics Course
for Teachers in July 2010.
We have gradually expanded the set of numbers we use: first from finger
counting to the whole set of positive integers, then to positive rationals, ir-
rational reals, negatives and finally to complex numbers. It has not always
been easy to accept new numbers. Negative numbers were rejected for cen-
turies, and complex numbers, the square roots of negative numbers, were
considered impossible.
Complex numbers behave like ordinary numbers. You can add, subtract,
multiply and divide them, and on top of that, do things which you cannot
do with real numbers. Today, complex numbers have many important ap-
plications in mathematics and physics, and scientists could not live without
them.
What if we take the next step? What comes after the complex numbers?
Is there a bigger number system that has the same nice properties as the real
numbers and the complex numbers?
The answer is yes. In fact, there are two (and only two) bigger number
systems that resemble real and complex numbers, and their discovery has
been almost as dramatic as that of the complex numbers.

1 Complex numbers
Complex numbers where discovered in the 16th century when Italian math-
ematicians tried to find a general solution to the cubic equation

x3 + ax2 + bx + c = 0.

1
 At that time, mathematicians did not publish their results but kept them
secret. They made their living by challenging each other to public contests of
problem solving in which the winner got money and fame. In these contests
it was useful to know things other people did not know. Also the formula for
solving cubic equations, or at least a part of it, was first kept secret.
The mathematicians of the time did not like negative numbers because
to them they had no meaning. What does minus three potatoes mean? The
square root of a negative number was even worse. It was impossible. These
numbers were referred as fictitious, absurd or false.
However, mathematicians noticed that sometimes when they used the
formula for the cubic equation, they had square roots of negative numbers
in the intermediary steps. When they were brave enough to treat these as
ordinary numbers for which the familiar laws of arithmetic hold, they noticed
that despite the impossible intermediary steps, the resulting root was in some
cases a real number.
Even though mathematician started to use the square roots of negative
numbers, they did not know what their geometrical interpretation was. It
was not before the 19th century before mathematicians realised how to treat
them as points of the so-called complex plane.
Definition 1. The set of complex numbers consists
√ of elements of the form
a + bi, where a and b are real numbers and i = −1.
The rules of arithmetic are defined as follows:
Addition: (a + bi) + (c + di) = (a + c) + (b + d)i
Subtraction: (a + bi) − (c + di) = (a − c) + (b − c)i
Multiplication: (a + bi) · (c + di) = (ac − bd) + (ad + bc)i
Division:
a + bi (a + bi)(c − di) ac + bd −ad + bc
= = 2 + 2 i.
c + di (c + di)(c − di) c + d2 c + d2

The Irish mathematician William Hamilton noticed that the complex

number a + bi can be written as a pair of real numbers (a, b) – or, a vector.
For these vectors it is possible to define the scalar multiplication by real
numbers, that is,
r · (a, b) = (ra, rb)

2
for all r, a, b ∈ R. This means that the set of complex numbers is just the
real plane R2 with multiplication defined in quite a weird way:

(a, b) · (c, d) = (ac − bd, ad + bc).

For each complex number a+bi, one can define its norm or length |a+bi|,
which is just
√ the length of the corresponding vector. In other words, we have
|a + bi| = a2 + b2 . The norm is compatible with multiplication:

|z1 · z2 | = |z1 | · |z2 |.

The complex conjugate of the complex number a + bi is a + bi = a − bi.

The rule for dividing complex numbers can now be expressed using lengths
and complex conjugates:
z1 1
= · z1 · z2 ,
z2 |z2 |2
where z1 , z2 ∈ C.
All these properties make complex numbers a very special number system,
a normed division algebra. Complex numbers have a norm (or length), and
one can divide complex numbers. We will now start to look for other normed
division algebras, that is, other number systems that have similar properties
as complex numbers.

2 Normed division algebras

This section is for those who want something more precise, and it can be
skipped.
A normed division algebra over the real numbers is the space Rm with
multiplication and division. (The space Rm consists of finite sequences of

3
m real numbers x = (x1 , x2 , . . . , xm ).) The addition, multiplication by real
numbers and the norm (or length) of a vector are defined as usual, and all
these operations behave in a nice way. In other words, in a normed division
algebra the following rules hold:

(x + y) · z = x · z + y · z,
x · (y + z) = x · y + x · z,
(ax) · (by) = (ab) · (xy),

where x, y and z are elements of Rm and a and b are real numbers.

For every element u and any non-zero element v there exists precisely one
element x such that u = v · x and precisely one element y such that u = y · v.
This means that we can divide elements of a normed division algebra.
Each element x of a normed division algebra has norm (or length) |x|,
which is a real number that satisfies the rule

|xy| = |x||y|.

Real and complex numbers are normed division algebras, and we will now
define two other division algebras. The aim is to find all such structures.

3 Higher dimensions
Now we have two nice number systems, R and C, in which it is possible
to add, subtract, multiply and divide. Also, in both cases we can define
the norm of an element. (In the case of real numbers the norm is just the
absolute value.)
The complex numbers are a 2-dimensional structure because every com-
plex number can be written as a sequence of two real numbers. The set of
real numbers is of course a 1-dimensional structure because every real number
can be written as a sequence of one real number.
Now that we have a 1-dimensional and a 2-dimensional number system,
it is natural to ask whether there is a 3-dimensional number system that has
the same nice properties as the real numbers and complex numbers?
In a case of an n-dimensional structure, it is easy to find out how to add
two things together, define the norm or multiply elements with real numbers.
It can be done in exactly the same way as in the case of complex numbers.

4
 The only difficult thing is the multiplication. You need to define it in
such a way that you can also divide. For example, one might think that the
obvious way of defining the multiplication of pairs would be (a, b) · (c, d) =
(a · c, b · d). However, in this case one cannot divide by pairs that have zero
in one component. For instance, we would have
 
(2, 1) 2 1
= , ,
(1, 0) 1 0

which is impossible. This explains why the multiplication of complex num-

bers is defined in such a complicated way.
Hamilton had realised how to treat complex numbers as pairs of real
numbers, and after that he wanted to find triples that would behave in a
similar manner than complex numbers.
Hamilton tried very hard for eight years, but could not define the multipli-
cation for his triples. In 1843 he finally realised that he actually needed four
dimensions. Hamilton was so thrilled about his discovery that he carved
the rules of multiplication of his 4-dimensional structure into a stone of
Brougham Bridge in Dublin.

4 Quaternions H
The structure Hamilton came up with is called quaternions (or Hamilto-
nians). One takes three square roots of −1: i, j and k. All the possible
combinations of 1, i, j, and k form the set of quaternions.

Definition 2. Quaternions are of the form a + bi + cj + dk, where a, b, c, d

are real numbers and i2 = j 2 = k 2 = −1.

Each quaternion a + bi + cj + dk can be written as a list of four real

numbers (a, b, c, d). This means that the quaternions are a 4-dimensional
structure.
As mentioned above, addition, multiplication by real numbers and norm
are easy to define. They are defined in the same way as in the case of complex
numbers:

a) (a+bi+cj+dk)+(a0 +b0 i+c0 j+d0 k) = (a+a0 )+(b+b0 )i+(c+c0 )j+(d+d0 )k

b) r · (a + bi + cj + dk) = ra + (rb)i + (rc)j + (rd)k, where r ∈ R

5
 √
c) |a + bi + cj + dk| = a2 + b2 + c2 + d2 .
The multiplication table is defined by the famous equations of Hamilton:

i2 = j 2 = k 2 = −1,
ij = k, jk = i, ki = j,
ji = −k, jk = −i, ik = −j.

The rules of multiplication can be visualised with the following picture in

which the arrow indicates the sign of the product.

Using these equations, one can calculate the product of two arbitrary
quaternions. For example, we have

(2 + 3i) · (3i − j) = 6i − 2j − 9 − 3k

and

(1 − 4j + 3k) · (−2i + j + 2k)

= −2i + j + 2k − 8k + 4 − 8i − 6j − 3i − 6
= −2 − 13i − 5j − 6k.

One can also divide quaternions by using norms and conjugates just as in
the case of complex numbers. The conjugate of the quaternion a+bi+cj +dk
is a + bi + cj + dk = a − bi − cj − dk. Now we have
x 1
= 2 ·x·y
y |x|
for all quaternions x and y.
Notice that unlike the multiplication of real and complex numbers, the
multiplication of quaternions is not commutative. In other words, in a prod-
uct you cannot change the order of the factors. For example, we have ij = k
but ji = −k.

6
 Hamilton was obsessed by quaternions, and devoted the rest of his life
to them. For a while the quaternions were very popular, and much of what
is now done with three-dimensional vectors was then done with quaternions.
Not everybody thought it was a good thing, and there was a fierce bat-
tle between those who wanted to use vectors and those who wanted to use
quaternions. For example, the great physicists Kelvin and Heaviside wrote
some devastating attacks against quaternions. It was the quaternions that
finally lost the battle.

5 Even higher dimensions

Hamilton’s friend John Graves was interested in quaternions, and thought
that there might be bigger structures with same kind of miraculous proper-
ties. Indeed, only a couple of months after Hamiltons great discovery, Graves
constructed an interesting 8-dimensional number system. Graves sent letters
describing his results to Hamilton who was impressed, and promised to pub-
lish Graves’ discovery.
However, Hamilton was very busy, and did not publish the results right
away. In 1845, young Englishman Arthur Cayley came up with similar
ideas, and published a paper in which he gave a description of the same
8-dimensional number system Graves had discovered earlied. Graves was of
course upset, and pointed out that he had known about the structure since
1843. But it was too late, and even though it was admitted that Graves
had invented the number system first, people had already started to call it
Cayley numbers.

6 Octonions O
Nowadays the 8-dimensional number system found by Graves and Cayley is
called octonions. This time one takes seven square roots of −1: i0 , i1 , . . . i6 .
All the possible combinations of 1 and these roots form the set of octonions.

Definition 3. Octonions are of the form

a + a0 i0 + · · · + a6 i6 ,

where a, a0 , . . . , a6 are real numbers and i2t = −1 for all t.

7
 Each octonion a + a0 i0 + · · · + a6 i6 can be written as a list of eight real
numbers (a, a0 , . . . a6 ). This means that the octonions are an 8-dimensional
structure.
Again, addition, multiplication by real numbers and norm are defined in
the familiar way:

addition:

(a + a0 i0 + · · · + a6 i6 ) + (a0 + a0 i00 + · · · + a06 i6 )

= (a + a0 ) + (a0 + a00 )i0 + · · · + (a6 + a06 )i6

multiplication by real numbers:

r · (a + a0 i0 + · · · + a6 i6 ) = ra + (ra0 )i0 + · · · + (ra6 )i6 , r ∈ R

norm: q
|a + a0 i0 + · · · + a6 i6 | = a2 + a20 + · · · + a26

To define the multiplication table, it is enough to do it for the octonions

i0 , i1 , . . . i6 . The rules of multiplication are described in the picture below.
The product of two octonions is the third octonion that is on the same
line. The arrow indicates what the sign of the product is. For example, the
octonions i1 , i2 and i4 are on the same line, so i1 i2 = i4 , i2 i1 = −i4 , i1 i4 = −i2
and so on.

Notice that every line in the diagram behaves like the quaternions i, j
and k.

8
 There is also another way of expressing the rule of multiplication:

the octonions it , it+1 , it+3 behave as the quaternions i, j, k.

The subscripts are understood modulo 7. For example, consider the product
i5 · i6 . Here t equals 5, so t + 3 equals 8. Because 8 − 7 = 1, the product
is equal to i1 . The triples (it , it+1 , it+3 ) correspond to the lines of the above
diagram.
Octonions can be divided by using norms and conjugates in the same way
as in the case of complex numbers and quaternions.
Notice that unlike multiplication of real numbers, complex numbers and
quaternions, the multiplication of octonions is not associative. In other
words, the place of the brackets in the octonion multiplication matters. For
example, we have (i0 i1 )i2 = i3 i2 = −i5 but i0 (i1 i2 ) = i0 i4 = i5 .

7 Other dimensions
Now we have four structures of four different dimensionalities:
1) Real numbers – dimension 1
2) Complex numbers – dimension 2
3) Quaternions – dimension 4
4) Octonions – dimension 8
What about other dimensions? Graves considered the idea of a general
theory of 2m -ions, but failed to construct even a 16-dimensional number
system in which it would be possible to divide. In other words, he could not
construct any other normed division algebras. Graves started to think that
it might be impossible, and he was right.
Every number system in our list can in some sense be built from the
real numbers. For example C = R + Ri. Therefore the dimension of C is
the dimension of R plus the dimension of R, that is, twice the dimension
of R. The quaternions, on the other hand, are built from complex numbers
(H = C + Cj). The dimension of quaternions is twice the dimension of C.
Finally, the octonions are built from quaternions, and therefore the dimension
of octonions is twice the dimension of the quaternions. This way we can
obtain new number systems by doubling the old ones.

9
 What happens after octonions? If we double octonions, we obtain a
16-dimensional set, but in this new set it is not possible to divide. Each
number system in our list is contained in the next one. When passing from
a number system to the next one, we always lose something, and this is why
the doubling process cannot be continued after octonions. In some sense,
octonions are too weird for doubling.
What kind of things do we lose? First we lose trivial conjugation. For
real numbers the complex conjugation a + bi 7→ a − bi does not do anything,
but for complex numbers it does. In other words, for the complex numbers
complex conjugation is non-trivial. Then we lose commutativity because in
quaternion multiplication the order of the factors matters. Finally, we lose
associativity because in octonion multiplication the place of the brackets
matters. It can be shown that it is the non-associativity of octonions that
makes dividing in the 16-dimensional set impossible. Therefore it is possible
to divide only in dimensions 1, 2, 4 and 8.

8 Literature
• John Baez: The octonions
http://math.ucr.edu/home/baez/octonions/

• Helen Joyce: Curious quaternions

http://plus.maths.org/issue32/features/baez/

• Helen Joyce: Ubiquitous octonions

http://plus.maths.org/issue33/features/baez/index.html
√
• Paul Nahin: An imaginary tale: The story of −1

10