Composite Quantum System

Postulate 4
Quantum systems 1,    , N are in states |ψ1 y,    , |ψN y. The state of the
composite system is |ψ1 y b    b |ψN y.

Example 1
System 1: |ψ1 y  ?12 p|00y |11yq
System 2: |ψ2 y  ?12 p|01y |10yq
Composite System: |ψ1 y b |ψ2 y  1
2
p|0001y |0010y |1101y |11110yq

Density Operator

Figure 1

Let someone measure a qubit, but does not reveal to us the classical output
of the measurement. In which state the qubit is now?
Answer: It is in a mixed state described by the set tp0  1{2, |0y; p1 
1{2, |1yu.
In general, if n qubits can be in state |ψi y P C2 , || |ψi y ||  1, with
n

probability pi , and p1 . . . pN  1, we say that the qubits are in the mixed

state tpi , |ψi y, i  1, . . . N u.

1
 Instead of working with tpi , |ψi y, i  1,    , N u we define density operator
¸
N
ρ pi |ψi yxψi |, p1  pN  1, || |ψiy ||  1, ı  1, . . . , N.

i 1

Example 2
 
ρ |0yx0| | 1yx1| 
1 1 1 1 0
2 2 2 0 1
Below we study the unitary evolution and measurement of mixed states
represented by density operators.

Unitary Evolution

Initial State New State

tpi, |ψiyu
ñ tpi, U |ψiyu
° p |ψ yxψ | ñ ° p U |ψ yxψ |U :
i1 i i i i i° i i
=U p i pi |ψi yxψi |qU :
 U ρU :
Measurements
Assume that we have 2n  2n projectors P1 ,    , Ps satisfying the conditoins:
1. P1 ... Ps  I2n ,
2. Pi Pj  δij Pi.
So, these projectors can be used for conducting a von Neumann measurement
of the n qubits.
Assume that our n qubits initially are in a mixed state

tpi, |ψiy, i  1, . . . , N u, (1)

which means that their density matrix is

¸
N
ρ pi |ψi yxψi |. (2)

i 1

2
 We would like to understand what are possible classical and quantum
outputs of the measurement and what are the corresponding probabilities.
Similar to the von Neumann measurement of qubits whose state is a
vector |v y P C2 , we again postulate that only one of the projectors, say Pm ,
n

acts on the qubits as a result of the measurement.

Since our qubits are in the mixed state (1), we analyse the measurement
considering the following two dependent random variables. The first ran-
dom variable is the quantum state |ψ y defined by the probabilities Prp|ψ y 
|ψiyq  pi. The second random variable is the projector operator P that acts
on qubits as a result of the measurement. P is defined by the conditional
probabilities

PrpP  Pm | |ψ y  |ψi yq
 PrpPm act on the qubits | qubits are in the state |ψiyq
This probability, according to the definition of the von Neumann measure-
ment considered in the previous lecture, is

PrpP  Pm| |ψy  |ψiyq  xψi|Pm|ψiy. (3)

Assuming that the qubits are effected by Pm , we want to find the condi-
tional probability that Pm acted on the state |ψi y. In other words, we want
to find the conditional probability

Prp|ψ y  |ψi y|P  Pmq. (4)

Note that if Pm acted on qubits in |ψi y, we obtain the state

|ψimy  a Pm|ψiy (5)

xψi|Pm|ψiy
So, we have
Prp|ψim y|P q  Prp|ψi y  |ψi y|P q. (6)
In order of finding this probability, we use Bayes’ rule

Prp|ψ y  |ψi y, P  Pm q
 Prp|ψy  |ψiyq  PrpP  Pm| |ψy  |ψiyq
 PrpP  Pmq  Prp|ψy  |ψiy|P  Pmq. (7)

3
Note that in the above expression we have
Prp|ψ y  |ψi yq  pi (8)
PrpP  Pm ||ψ y  |ψi yq  xψi |Pm |ψi y, (9)
where pi are the same probabilities as in (1). Next,
¸
PrpP  Pm q  Prp|ψ y  |ψi yq PrpP  Pm| |ψy  |ψiyq
¸
i

 pi xψi |Pm |ψi y

¸i ¸
 pi Trpxψi |Pm |ψi yq  pi TrpPm |ψi yxψi |q
i
¸ i

TrpPm pi |ψi yxψi |q  TrpPm ρq. (10)

i

Combining (6) - (10), we obtain

pi xψi |Pm |ψi y
Prp|ψim y|P q 
TrpPm ρq
. (11)

Now we have everything for finding the density operator that we get after
the measurement condition that the projector Pm effected the n qubits:
¸
ρ1  Prp|ψim y|P q|ψim yxψim |

¸
i

 pixTrψip|PPmρ|ψq iy Pmxψ|ψ|iPyxψ|ψi|Pym
i m
¸ i m i
 TrpP1 ρq Pm pi|ψiyxψi| Pm
m i

 Tr
Pm ρPm
pP ρq .
m

Summarising these derivations, we conclude that as a result of the mea-

surement we get the following classical and quantum outputs with the fol-
lowing probability

Classical Output Quantum Output Probability

m ρm  TrP pPρP ρq
m
m
m
ppmq  TrpPm ρq

4
 If, for some reason, we do not have an access to the classical output of the
measurement, and therefore we do not know which Pm acted on the qubits,
then the best what we can say is that qubits after the measurement are in
the mixed state
¸ ¸
ρ1  ppmqρm  TrpPm ρq
Pm ρPm
TrpPm ρq
¸
m m

 Pm ρPm .
m

Properties of Density Operators

Definition 1 Quantum state |ψ y P C2 , || |ψy ||  1, is a pure state. In this
n

case ρ  |ψ yxψ |.

Definition 2 If ρ  |ψyxψ| for any |ψy P C2 , || |ψy ||  1, then this is a

n

mixed state.

We will prove below that

Trpρ2 q  1 ñ Pure State

Trpρ2 q 1 ñ Mixed State

Any density operator has the form

¸
N
ρ pi |ψi yxψi |, pi ¡ 0, p1    pN  1, |ψiy are pure q. states. (12)

i 1

Theorem 1 ρ is a density operator, i.e. it can be represented in form (12),

iff

1. Trpρq  1

2. ρ is a positive semi-definite operator, i.e., ρ:  ρ and xv|ρ|vy ¥ 0 for any |vy.

Proof

5
 1.
¸ ¸
Trpρq  Tr pi |φi yxφi |  pi Trp|φi yxφi |q
¸ i
¸ i

 pi Trpxloomoon
ψi |ψi yq  pi  1.
i i
1

2. ¸ ¸
xv|ρ|vy  xvomo
pi lo |φoiyn loxφomo
i |voy
n  pi |a2i | ¥ 0.
a
i i
ai
i

Conversely, let ρ satisfies (1) and (2). ρ is Hermitian, hence using the spectral
decomposition, which we formulated a couple lectures ago, we get
¸
ρ λj |φj yxφj |, xφi |φj y  0, i  j, || |φj y ||  1. (13)
j

¥ 0. Thus
Since ρ is positive semi-definite, we have λj
¸ ¸ ¸
Trpρq  1 ñ Trpρq  λj Trp|ψj yxψj |q  ψj |ψj yq 
λj Trpxloomoon λj  1.
j j j
1

If we assign pj =λj and |ψj y  |φj y, we get ρ in the form (12).

Note that if qubits are in a pure state |φy then their density operator is
ρ  |φyxφ|. (14)
Theorem 2 If Tr(ρ2 q  1 then ρ is a pure state, i.e., it has form (14). If
Tr(ρ2 q 1 then ρ is a mixed state, i.e., it has form (12) with N ¥ 2.
Proof Using (13) we get
¸
N ¸
N
Trpρ 2
q λi λj Trp|ψj y loomoon
xψj |ψiyxψi|q
 
i 1j 1
δij

¸
N
 λ2i Trpxloomoon
ψi |ψi yq

i 1
1
¸
N ¸
N
 λ2i 1 if N ¥2 and λi  1.
i i 1 

6
 It is natural to ask whether different sets tpi , |φi y, i  1,    , tu and
tp1i, |ψiy, i  1,    , su can result in the same density operator ρ. As the
following example shows, the answer to this question is positive.
Example 3

tp1  34 , |φ1y  |0y; p2  41 , |φ2y  |1yu, (15)

 
3{4 0
ρ  |0yx0| |1yx1|      0 1{4 ;
3 1
4 4
c c c c
tp 1  1
, |ψ1 y 
3
| 0y
1
| 1y; p12  , |ψ2 y 
1 3
|0y
1
|1yu, (16)
 2 
1
2 4 4 4 4
3{4 0
ρ1 12 |ψ1yxψ1| |ψ2yxψ2|      0 1{4 .
1
2
ñ ρ  ρ1 .
Note that sets (15) and (16) have the same “rights”. One is not better than
the other.
We can characterize sets that give the same ρ. Let
¸s ¸t
ρ1  pi |φi yxφi |, ρ2  qj |ψj yxψj |.

i 1 
j 1

If say s ¡ t, then
¸s
ρ2  qj |ψj yxψj |, with qj  0, j ¡ t.

j 1
So we may assume s  t.
Theorem 3 (Unitary Freedom) ρ1  ρ2 iff
r p |φ y . . . p |φ ys  r?q |ψ y . . . ?q |ψ ysU,
?
1 1
?
s s 1 1 s s

for some s  s unitary U .

Proof is omitted.
Example 4 In the previous example we can take
 
U?
1 1 1
2 1
1 (17)

to move one ensemble into another one.

7
1 General Form of Quantum Postulates
Below we formulate a more general form of quantum postulates, compared
to the ones we formulated before, using the density operators.

ˆ Postulate 1. The state of a closed quantum system of n qubits is a

2n  2n density operator ρ.

ˆ Postulate 2. The evolution of a closed quantum system is unitary

ρ1  U ρU :
where ρ1 is the new state and ρ is the old state.

ˆ Postulate 3. Quantum measurement is described by a set of projectors

P1 ,    , Pt such that

1. Pi Pj  0 if i  j
2. P1    Pt  I2
n

Classical Output Quantum Output Probability

m Pm ρPm
ρm = Tr pPm ρq Pm  TrpPmρq

ˆ Postulate 4. If quantum systems 1,    , N have states ρ1 ,    , ρN then

the state of the composite system is

ρ  ρ1 b    b ρN .