Quantum Computing

Lecture 3

Postulates of Quantum Mechanics

Maris Ozols

What is quantum mechanics?

Quantum mechanics is a branch of physics that describes the behaviour of

systems, such as atoms and photons, whose states admit superpositions.

It is a framework onto which other physical theories are built upon. For
example, quantum field theories such as quantum electrodynamics and
quantum chromodynamics.

The central topic of this lecture is a mathematical formulation of

quantum mechanics consisting of four postulates.

This lecture is based on Section 2.2 of the book by Nielsen & Chuang.
 What are the four postulates about?

Open vs closed systems

Closed system is an ideal physical system that does not interact at all
with its environment. An open system does interact with its environment.

Postulates
They specify a general framework for describing the behaviour of a
physical system:

1. Statics (state space): describes the state of a closed system

2. Dynamics: describes the evolution of a closed system
3. Measurement: describes how information is extracted from a
closed system via interactions with an external system
4. Composite systems: describes the state of a composite system in
terms of its component parts

First Postulate
The state space of any closed physical system is a complex vector space.
At any given point in time, the system is completely described by a state
vector, which is a unit vector in its state space.

Note: Quantum mechanics does not prescribe what the state space of a
particular physical system is, this is determined by more specific theories.
Any physical system whose state space can be described by C2 can serve
as an implementation of a qubit.
Examples:
• spin of an electron
• polarization of a photon
• current in a superconducting circuit

Some systems may require an infinite-dimensional Hilbert space as their

state space. However, for the purpose of this course we always assume
that our systems are finite-dimensional.
 Second Postulate
The continuous-time evolution of a closed quantum system is described
by the Schrödinger equation:
d
i |ψ(t)i = H|ψ(t)i
dt
where H is a fixed Hermitian operator known as the Hamiltonian.

By solving this differential equation one gets:

|ψ(t)i = U (t)|ψ(0)i where U (t) = exp(−iHt)

and |ψ(0)i is the state at t = 0. One can check that U (t) is unitary.
While some models (such as adiabatic quantum computing) allow for
continuous-time evolution, we consider only discrete computational steps.

The discrete-time evolution of a closed quantum system is described by a

unitary transformation U :
|ψ 0 i = U |ψi

Expressing a state in any basis

Any state |ψi ∈ Cn can be expressed in the standard basis as follows:

 
α1 n n
 ..  X X
|ψi =  .  = αi |ii = hi|ψi|ii
αn i=1 i=1

where αi = hi|ψi is the i-th coordinate of |ψi (in the standard basis).

Similarly, if {|u1 i, . . . , |un i} is any other orthonormal basis of Cn , we can

express |ψi in this basis as follows:
n
X
|ψi = hui |ψi|ui i
i=1

where hui |ψi is the i-th coordinate of |ψi in the basis {|u1 i, . . . , |un i}.
 Unitary change of basis
Let {|u1 i, . . . , |un i} be some orthonormal basis of Cn . Then we can
express any |ψi ∈ Cn in two different ways:
n
X n
X
|ψi = αi |ii = βj |uj i
i=1 j=1

for some coordinates αi , βj ∈ C. How are αi and βj related?

If we left-multiply both sides by hi|, we get

n
X n
X
αi = βj hi|uj i = Mij βj
j=1 j=1

where Mij = hi|uj i. Since this holds for every i,

   
α1 β1
 ..   . 
 .  = M  .. 
αn βn

Unitary change of basis (continued)

If we left-multiply by M −1 , we get
   
α1 β1
M −1  ...  =  ... 
   

αn βn

But we could have done the whole calculation the other way and got
   
α1 β1
N  ...  =  ... 
   

αn βn

where Nij = hui |ji = hj|ui i = Mji . We conclude that M −1 = N = M † .

In particular, M M † = M † M = I so M is unitary! The same holds for N .

 Unitary change of basis (continued 2)
Since N allows us to compute from αi the new coordinates βj , we can
use it to convert any vector from the standard basis {|1i, . . . , |ni} to the
new basis {|u1 i, . . . , |un i}.

Recall that Nij = hui |ji, so we can write N explicitly as follows:

n
X n
X n
X n
X n
X
N= Nij |iihj| = |iihui |jihj| = |iihui | |jihj| = |iihui |
i,j=1 i,j=1 i=1 j=1 i=1
Pn
where we recalled that j=1 |jihj| = I, the identity matrix.

Summary: To go from the standard basis |ii to another orthonormal

basis |ui i, we use the following unitary change of basis transformation:
n
X
U= |iihui |
i=1

Expressing a matrix in a different basis

Assume we are given the entries Aij = hi|A|ji of some matrix

A ∈ Mn,n (C). How can we express the same matrix in a different basis?

That is, how do we compute a matrix B such that Bij = hui |A|uj i
where {|u1 i, . . . , |un i} is some orthonormal basis?

Note that
n
X n
X
B= Bij |iihj| = |iihui |A|uj ihj| = U AU †
i,j=1 i,j=1
Pn
where U = i=1 |iihui | is the basis change unitary!

Summary: If U is the change of basis from |ii to |ui i then U AU † is the

matrix A expressed in the new basis |ui i.
 Pauli gates
A particularly useful set of one-qubit unitaries are the Pauli gates:

I X Y Z

 
1 0
• The I gate: I = , I|0i = |0i, I|1i = |1i
0 1
 
0 1
• The X gate: X = , X|0i = |1i, X|1i = |0i
1 0
 
0 −i
• The Y gate: Y = , Y |0i = i|1i, Y |1i = −i|0i
i 0
 
1 0
• The Z gate: Z = , Z|0i = |0i, Z|1i = −|1i
0 −1

Note: Pauli matrices have lots of nice properties and are closely related
to quaternions: {I, iZ, iY, iX} ∼
= {1, i, j, k}.

Third Postulate
A measurement with input dimension n, output dimension m, and
outcome set S is a collection of |S| matrices of size m × n,
{Pk : k ∈ S} ⊂ Mm,n (C)
known as measurement operators, that satisfy the completeness relation
X †
Pk Pk = I n
k∈S

If the system is in state |ψi ∈ Cn before the measurement, the

probability of outcome k ∈ S and the corresponding post-measurement
state |ψk i ∈ Cm is
Pk |ψi
p(k) = hψ|Pk† Pk |ψi = kPk |ψik
2
|ψk i = q
hψ|Pk† Pk |ψi

Probabilities of all outcomes add up to 1:

hψ|Pk† Pk |ψi = hψ|In |ψi = 1
X X
p(k) =
k∈S k∈S
 Orthogonal measurement
An orthogonal measurement is a measurement whose measurement
operators are projectors
Pk = |uk ihuk |
where {|u1 i, . . . , |un i} ⊂ Cn is an orthonormal basis. When measuring
state |ψi ∈ Cn , the probability of outcome k ∈ {1, . . . , n} and the
corresponding post-measurement state is
2
p(k) = |huk |ψi| |ψk i = |uk i

The computational or standard basis measurement corresponds to the

case when |uk i = |ki.
Example: When |ψi = α|0i + β|1i is measured in the standard basis,
2
P0 = |0ih0|, p(0) = |α| |ψ0 i = |0i
2
P1 = |1ih1|, p(1) = |β| |ψ1 i = |1i

Relative phase matters

Recall these two states from the first lecture:
1 1
|+i = √ (|0i + |1i) |−i = √ (|0i − |1i)
2 2
They cannot be distinguished by measuring in the computational basis
since both outcomes occur with probability 1/2.
However, these states themselves form an orthonormal basis {|+i, |−i},
the Hadamard basis. When measuring |+i in this basis, the probabilities
are
2 2
p(+) = |h+|+i| = 1 p(−) = |h−|+i| = 0

so we always get the outcome “+”. Similarly, when |−i is measured in

this basis we always get the outcome “−”.
While the standard basis measurement produces a uniformly random
outcome and thus gives no information about which of the two states we
have, the Hadamard basis measurement identifies the state perfectly!
 Haidinger’s brush

Source: Wikipedia

Fourth Postulate

The state space of a composite physical system is the tensor product of

the state spaces of the individual component physical systems. If one
component is in state |ψ1 i and a second component is in state |ψ2 i, the
state of the combined system is |ψ1 i ⊗ |ψ2 i.

If the joint state of a system is |ψ1 i ⊗ |ψ2 i and the first party applies U ,
the new state is

(U ⊗ I) ⊗ (|ψ1 i ⊗ |ψ2 i) = (U |ψ1 i) ⊗ |ψ2 i

This is the same as the combined state of U |ψ1 i and |ψ2 i.

However, not all states of a combined system can be separated into the
tensor product of states of the individual components. . .
 Why tensor product?
Imagine you have two random coins:
   
p0 q
P = Q= 0
p1 q1

What is their joint probability distribution?

    
00 : p0 q0 q
p0 0     
01 : p0 q1  
   q1  p0 q0
=   = ⊗ =P ⊗Q
10 : p1 q0   q0  p1 q1
11 : p1 q1 p1
q1

Similarly, if you have to qubit states

   
α0 β0
|ψi = |ϕi =
α1 β1

their joint state is |ψi ⊗ |ϕi. Note that k|ψi ⊗ |ϕik = k|ψikk|ϕik = 1.

Computational basis: notation

   
α0 |0i β0 |0i
|ψi = |ϕi =
α1 |1i β1 |1i


    α0 β 0 |00i
α0 β0 α0 β1  |01i
|ψi ⊗ |ϕi = ⊗ = 
α1 β1  α1 β0  |10i
α1 β1 |11i

Standard basis notation for the joint system: |ii ⊗ |ji ≡ |i, ji ≡ |iji.
For example:
       
1 0 0 0
0 1 0 0
|00i = 
0
 |01i = 
0
 |10i = 
1
 |11i =  0


0 0 0 1
 Product and entangled states
A state |Ψi ∈ Cn ⊗ Cm of a combined system is product if it can be
expressed as |Ψi = |ψ1 i ⊗ |ψ2 i for some |ψ1 i ∈ Cn and |ψ2 i ∈ Cm .
Otherwise it is called entangled.

Example: This two-qubit state is a product state:

1 1 1
(|00i + |01i + |10i + |11i) = √ (|0i + |1i) ⊗ √ (|0i + |1i)
2 2 2

Example: Neither of the following two-qubit states can be written as a

product of single-qubit states, hence they are both entangled:
1 1
√ (|10i + |01i) and √ (|00i + |11i)
2 2

Note: Physical separation does not imply that the joint state must be
product! Just like two distant random coins can still be correlated, two
physically separated particles can also be entangled.

How to measure only one of two qubits?

Given a state |ψi ∈ C2 ⊗ C2 , how do we measure only the first qubit?
We tensor the desired measurement operators with I! For example, if we
want to measure the first qubit in the standard basis, we take

Pk = |kihk| ⊗ I = (|ki ⊗ I)(hk| ⊗ I)

Then the probability to get outcome k is

2
p(k) = k(hk| ⊗ I)|ψik

and the post-measurement state of the two qubits is

(hk| ⊗ I)|ψi
|ψk i = |ki ⊗
k(hk| ⊗ I)|ψik

If we do not want to keep the first qubit around after the measurement
and want to discard altogether, we can simply take Pk0 = hk| ⊗ I.
 Summary

• Postulate 1: A closed system is described by a unit vector in a

complex vector space.
• Postulate 2: The evolution of a closed system in a fixed time
interval is described by a unitary transformation.
• Postulate 3: If a closed system is in state |ψi and we measure it in
an orthonormal basis {|u1 i, . . . , |un i}, we get outcome k with
2
probability |huk |ψi| and the system is now in the state |uk i.
• Postulate 4: The state space of a composite system is the tensor
product of the state spaces of its components.
Pn
• Expanding a state in any basis: |ψi = i=1 hui |ψi|ui i
Pn
• Change of basis: go from |ii to |ui i using U = i=1 |iihui |
†
• Matrix in a different basis: if A is in basis |ii then U AU is in |ui i
• Product state: |ψ1 i ⊗ |ψ2 i
√
• Entangled state: not product, e.g., (|00i + |11i)/ 2