Notes 8.370/18.

435 Fall 2022

Lecture 2 Prof. Peter Shor
Today, I’ll start explaining the quantum mechanics you will need for this course.
I’ll start with a statement of what physicists call the Superposition Principle or the
Linearity Principle of quantum mechanics. The following statement is expressed close
to the way you’ll see it physics literature.
The Superposition Principle
Suppose | Ay and | By are two perfectly distinguishable quantum states of a quantum
system, and α, β P C with |α2 ` β 2 | “ 1. Then

α | Ay ` β | By

is a valid state of the quantum system.

Here | Ay and | By are notation which physicists use for quantum states. It simply
means that A and B can be described by complex column vectors.
What perfectly distinguishable means here is that there is some experiment that can
distinguish between these two states, at least theoretically (all experiments will have
some amount of noise in practice).
There is a second part of this principle—if the experiment distinguishing | Ay from
| By is applied to the state α | Ay ` β | By, then the probability of getting the answer
| Ay is |α|2 and the probability of | By is |β 2 |.
How can we rewrite this principle using mathematics?
First, the fact that you can add states α | Ay ` β | By means that the quantum state
space is a complex vector space. What also follows, but is harder to see, is that a
quantum state is a unit vector in the vector space. Two states are orthogonal if and only
if they are perfectly distiguishable.
In this course, we will be talking about quantum computation. Let us recall that
classical computers work on bits, which are systems that are in one of two states.
Traditionally, we call the values of these bits 0 and 1, independent of the details of
their physical representation. For example, on an integrated circuit, a higher voltage
might represent a 1 and a low voltage might represent a 0. Or in a magnetic memory, a
region where the magnetic north is oriented upward might be a 0 and where it’s oriented
downward might be a 1.
The same thing is true of qubits. A qubit is a quantum system that can only take on
two distinguishable values. But these values might be the polarization of a photon, or
the spin of an electron, or the ground state and an excited state of an ion, or a clockwise
and a counterclockwise rotation of a current in a superconductor. We will give two
examples in this lecture; the first is the polarization of a photon, and the second is the
spin of a spin- 21 particle (such as an electron, a proton, or a neutron). Polarization is
probably the example most people are familiar with, so we start with that.
You may be familiar with polarized sunglasses. A polarizing filter only lets ver-
tically polarized (or horizontally polarized) photons pass through. Because sun glare
tends to be horizontally polarized, if you use vertical polarizing filters as lenses, they
screen out most the glare, but let through most of the rest of the light, and so reduce

1
glare. Of course, you can turn these sunglasses 45˝ , so they only let right diagonally
polarized light through. How do we explain this in terms of the quantum state space?
It is the case that horizontal and vertical polarization are two distinguishable vec-
tors. This means that other polarizations are linear combinations (called superposi-
tions) of these.
Now, we need to introduce a little physics notation. Physicists like to represent
iquantum states using a funny notation they call “kets”, so a quantum state v would be
represented as | vy. In mathematics notations, kets are column vectors. So | Øy and
| Ùy are the horizontal and vertical polarizations of light.
Because we’re doing quantum computation and working with qubits, we will rep-
resent these abstractly as | 0y and | 1y, using the convention that | 0y “ | Øy and
| 1y “ | Ùy. For example, if we have the vector
ˆ ? ˙
1{?2
v“ ,
i{ 2

then
1 i 1
| vy “ ? | 0y ` ? | 1y “ ? p| Øy ` i | Ùyq
2 2 2
Other polarizations are linear combinations (called superpositions) of | Øy and | Ùy.
For example,

|Õ
Öy “ ?1 p| Øy ` | Ùyq “ ?1 p| 0y ` | 1yq ,
2 2
|Ô
Œy “ ?1 p| Øy ´ | Ùyq “ ?1 p| 0y ´ | 1yq .
2 2

We will be using the states ?12 p| 0y ` | 1yq and ?12 p| 0y ´ | 1yq often enough that it
helps to have special names for these; we call these | `y and | ´y respectively.
Since the quantum state space is a complex vector space, we can also use complex
coefficients in our superposition. What does this give us? It turns out that these give
right and left circularly polarized light, | œy and | öy:

| œy “ ?1 p| Øy ` i | Ùyq ,
2
| öy “ ?1 p| Øy ´ i | Ùyq .
2

And if you use unequal coefficients in the above equations, you get elliptically polar-
ized light.
Also note that the choice of whether ?12 p| Øy ` | Ùyq is | Õ Ö y or | Ô
Πy is purely a
convention; I’ll try to stick with the conventions that the textbook uses in these notes.
There is another half of the ket notation: a row vector is represented by a bra: x¨ |.
By convention, xv | is the conjugate transpose of | vy, so if
ˆ ? ˙
1{?2
| vy is the column vector ,
i{ 2

then ´ ? ? ¯
xv | is the row vector 1{ 2, ´i{ 2 ,

2
Together, the “bracket” xv|wy is the inner product of xv | and | wy. Since quantum states
are unit vectors, and going from the ket to the bra involves a complex conjugation, we
get
xψ|ψy “ 1
for any quantum state | ψy. The notation xφ|ψy represents the inner product of two
states, and is called a bracket (and here you see the wordplay that Dirac used to come
up with “bra” and “ket”.
How can we change a quantum state? We will be doing quantum computation,
so we need to know how to change the state of our quantum computer. Physics says
that the evolution of an isolated quantum system is linear, and if we apply a linear
transformation to a vector, we have a matrix.
Suppose we have a matrix M that operates on a quantum state | Ψy. If this trans-
formation is something that can be implemented physically, the result of applying M
to a quantum state | ψy, which is M | ψy, has to be a unit vector, so we need

xψ | M : M | ψy “ 1 for all quantum states | ψy .

This implies that M : M “ I, the identity matrix. This is the condition for M being a
unitary matrix. Unitary matrices are essentially rotation matrices in complex space.
There are other conditions for a matrix U being unitary besides U : U “ I. A matrix
is unitary if and only if any of the following conditions hold:
• U : U “ I,
• U U : “ I,
• the columns of U are orthonormal vectors,

• the rows of U are orthonormal vectors.

There are three unitary matrices that operate on qubits (and thus are 2ˆ2 which are
very useful for quantum computing (and also for quantum physics in general). These
are ˆ ˙ ˆ ˙ ˆ ˙
0 1 0 ´i 1 0
σx “ σy “ σz “ .
1 0 i 0 0 ´1
These are called the Pauli matrices. A silly mnemonic for remembering which entry
has ´i in σy is that ´i is lighter than i, and thus floats to the top.
How do these matrices act on a qubit. Why are they labeled x, y, and z? To answer
this, we will use a different representation of a qubit, namely, the spin of a spin 12
particle. A spin 12 particle, such as an electron, only has two completely distinguishable
states of its spin, which we will take to be up and down. Of course, because of three-
dimensional symmetry, you can prepare a spin 12 particle with its spin axis pointing in
any direction. However, these other directions are linear combinations of spin up and
spin down. Namely, using the most common convention, Let us take up and down to
be the z-axis, and left and right to be the x-axis, so the y-axis goes into and out of the

3
board. Then we will let | 0y “ | Òy and | 1y “ | Óy. The standard convention is:

| Ñy “ ?1 p| Òy ` | Óyq ,
2
| Ðy “ ?1 p| Òy ´ | Óyq ,
2
| iny “ ?1 p| Òy ` i | Óyq ,
2
| outy “ ?1 p| Òy ´ i | Óyq ,
2

We can represent these states on the Bloch sphere (See Figure.) Here, each point
on the Bloch sphere represents a quantum state of a spin 21 particle that that has spin in
the corresponding direction.

ˆ ˙
0 1
What happens when we apply σx “ ? One can see that it takes the
ˆ ˙ ˆ ˙ 1 0
1 0
vector | 0y “ to | 1y “ and vice versa. It also takes the vector
ˆ ˙ 0 1 ˆ ˙ ˆ ˙
1 1 ´1
| `y “ ?12 to itself, and | ´y “ ?12 to ?12 “ ´ | ´y . Ex-
1 ´1 1
cept for the fact that | ´y is taken to ´ | ´y, this is a 180˝ rotation around the x-axis.
In fact, in some sense this really does represent a 180˝ rotation around the x-axis.
This is because multiplying like a unit complex vector like ´1, i, or eiθ does not change
the essential physical state — two quantum states | ψy and eiθ | ψy are indistinguishable
by any quantum measurement. We will see this when we get to measurements next
week. So the matrix σx represents a 180˝ rotation of the Bloch sphere around the x
axis. Similarly, the matrices σy and σz represent rotations of the Bloch sphere around
the y- and z-axes, respectively.
We will come back to the Bloch sphere later and tell you how to find the quantum
state pointing in an arbitrary direction along the Bloch sphere, as well as the unitary
matrix corresponding to an arbitrary rotation of the Bloch sphere.