Fernando Falceto

Departamento de Fsica Te
orica.
Universidad de Zaragoza

An introduction to
quantum circuits.
Martes cuantico, March 22, 2016

Why to study quantum circuits?

An interesting way of looking at quantum mechanics.

Miniaturization makes it necessary.

Interesting new algorithms.

New possibilities to simulate quantum systems.

2 / 29

Why to study quantum circuits?

An interesting way of looking at quantum mechanics.

Miniaturization makes it necessary.

Interesting new algorithms.

New possibilities to simulate quantum systems.

2 / 29

Why to study quantum circuits?

An interesting way of looking at quantum mechanics.

Miniaturization makes it necessary.

Interesting new algorithms.

New possibilities to simulate quantum systems.

2 / 29

Why to study quantum circuits?

An interesting way of looking at quantum mechanics.

Miniaturization makes it necessary.

Interesting new algorithms.

New possibilities to simulate quantum systems.

2 / 29

Why to study quantum circuits?

An interesting way of looking at quantum mechanics.

Miniaturization makes it necessary.

Interesting new algorithms.

New possibilities to simulate quantum systems.

2 / 29

Classical circuits.
INPUT

OUTPUT

x1
x2
x3
x4

z1

F
z2

x5

Variables xi , zj take values 1 (true) and 0 (false).

This circuit represents a boolean function
F : {0, 1}5 {0, 1}2

3 / 29

Classical circuits.
INPUT

OUTPUT

x1
x2

z1

x3
x4

z2

x5

Variables xi , zj take values 1 (true) and 0 (false).

This circuit represents a boolean function
F : {0, 1}5 {0, 1}2
Classical logic circuits are made of wires and gates.

3 / 29

Some classical gates.

x

1x

x
y

NOT

x
y

xy
AND

x y + xy

x
y

xy
XOR

OR

= y

1 xy

NAND

x y = x + y (mod 2).

4 / 29

Some classical circuits.

The full adder: adds two bits with carries.

5 / 29

Some classical circuits.

1968: SN74LS181, first Arithmetic Logic Unit in a single chip.

Able to performs 32 different operations with four-bit numbers.

6 / 29

Some classical circuits.

1968: SN74LS181, first Arithmetic Logic Unit in a single chip.

Able to performs 32 different operations with four-bit numbers.

6 / 29

Universal classical gates.

Which logic gates do we need to build any classical circuit?
Consider a general boolean function in n + 1 variables
f : {0, 1}n+1 {0, 1}
x0
x1
x2

f (x0 , x1 , x2 , . . . , xn )

xn
We call:

f0 (x1 , x2 , . . . , xn ) = f (0, x1 , x2 , . . . , xn )
f1 (x1 , x2 , . . . , xn ) = f (1, x1 , x2 , . . . , xn ).

Then...

7 / 29

Universal classical gates.

fx0 (x1 , . . . , xn ) = f (x0 , x1 , . . . , xn )

x0
x1
x2

f0 (x1 , . . . , xn )

xn
z

f1 (x1 , . . . , xn )

8 / 29

Universal classical gates.

fx0 (x1 , . . . , xn ) = f (x0 , x1 , . . . , xn )

x0
x1
x2

f0 (x1 , . . . , xn )

xn
z

f1 (x1 , . . . , xn )

8 / 29

Universal classical gates.

fx0 (x1 , . . . , xn ) = f (x0 , x1 , . . . , xn )

x0
x1
x2

f0 (x1 , . . . , xn )

xn
z

f1 (x1 , . . . , xn )

8 / 29

Universal classical gates.

fx0 (x1 , . . . , xn ) = f (x0 , x1 , . . . , xn )

x0
x1
x2

f0 (x1 , . . . , xn )

xn
z

f1 (x1 , . . . , xn )

8 / 29

Universal classical gates.

fx0 (x1 , . . . , xn ) = f (x0 , x1 , . . . , xn )

x0
x1
x2

f0 (x1 , . . . , xn )

xn
z

f1 (x1 , . . . , xn )

8 / 29

Universal classical gates.

fx0 (x1 , . . . , xn ) = f (x0 , x1 , . . . , xn )

x0
x1
x2

f0 (x1 , . . . , xn )

xn
z

f1 (x1 , . . . , xn )

8 / 29

Universal classical gates.

Then
,
NOT

,
AND

XOR

FANOUT

form a universal set of gates.

Every boolean function
f : {0, 1} {0, 1} {0, 1}
|
{z
}
n

can be iteratively constructed from these gates.

And, therefore, any F : {0, 1}n {0, 1}m .

9 / 29

Universal classical gates.

Then
,
NOT

,
AND

XOR

FANOUT

form a universal set of gates.

Every boolean function
f : {0, 1} {0, 1} {0, 1}
|
{z
}
n

can be iteratively constructed from these gates.

And, therefore, any F : {0, 1}n {0, 1}m .

9 / 29

Universal classical gates.

Then
,
NOT

,
AND

XOR

FANOUT

form a universal set of gates.

Every boolean function
f : {0, 1} {0, 1} {0, 1}
|
{z
}
n

can be iteratively constructed from these gates.

And, therefore, any F : {0, 1}n {0, 1}m .

9 / 29

Universal classical gates.

But:
=

=
NOT

AND

=
XOR

can be obtained combining only

,
NAND

FANOUT

They form a universal set of gates for classical circuits.

10 / 29

Universal classical gates.

But:
=

=
NOT

AND

=
XOR

can be obtained combining only

,
NAND
Peirce, 1881

FANOUT
Sheffer, 1913

They form a universal set of gates for classical circuits.

10 / 29

Programmable classical circuit

Design a circuit that can compute any f : {0, 1}n {0, 1}.
One would need 2n selection bits.
For n = 1

s1
s2

0
1

0,0
0
0
z=0

Selection bits: s1 , s2
0,1
1,0
0
1
1
0
z=x

z=x

1,1
1
1
z=1

11 / 29

Reversible logic circuits.

NAND gate is not reversible
F : {0, 1}n {0, 1}n
1. Landauer principle (1961): For every bit of information lose in a
circuit a minimum heat of kT ln 2 is disipated.

R. Dreschler, R. Willie. LNCS 7373 (2012)

A. B
erut et al. Nature 483 (2012)

2. Quantum circuits are reversible.

12 / 29

Reversible logic circuits.

Reversible gates:
x
x

NOT

11
00
00
11

00
11
000000000
111111111
1010 x
00
11
111111111
000000000
001010 yx x x
11
111111111
000000000

x1
x2
y

x
yx

CNOT

1 2

TOFFOLI

Invertible functions,
F : {0, 1}n {0, 1}n

.
.
.

.
.
.

= ..
.

t1

.
.
.

t2

. . .
. . .

tr

.
.
.

. . .

are a product of transpositions: F = tr . . . t2 t1 .

13 / 29

Reversible logic circuits.

0
1
1

1
0
1
1

1
0
1

0
1
1
0

Toffoli gate (+ ancilla bits) universal for

reversible classical circuits.
Toffoli, 1980

14 / 29

Quantum circuits
Qubit: Smallest (non trivial) quantum system.
Hilbert space C2 with computational basis {|0i , |1i}.
Quantum circuits: Unitary operators U : H H acting on
systems composed of several qubits, H = C2 C2 .

.
.
.

.
.
.

Quantum circuits are reversible.

No-cloning theorem: FANOUT gate

not allowed.

15 / 29

Some quantum gates

Single qubit gates

Multiple qubit gates

|xi

|yi

|yi

|xi

|xi

|xi

|yi

|y xi

|xi

|xi

SWAP

CNOT

CondU
|yi

Toffoli

U x |yi

|x1 i

|x1 i

|x2 i

|x2 i

|yi

|y x1 x2 i

16 / 29

Some quantum gates

Single qubit gates

Multiple qubit gates

|xi

|yi

|yi

|xi

|xi

|xi

|yi

|y xi

|xi

|xi

SWAP

CNOT

CondU
|yi

Toffoli

U x |yi

|x1 i

|x1 i

|x2 i

|x2 i

|yi

|y x1 x2 i

Measurements
|xi

x
U

Measurement

Classical control

Principle of deferred measuremet

16 / 29

Some useful relations

=
U =V2

Toffoli from CNOT!!!

U = ei AXBXC, with


0 1
ABC = 1, X =
1 0

17 / 29

Some useful relations

=
U =V2

Toffoli from CNOT!!!

U = ei AXBXC, with


0 1
ABC = 1, X =
1 0

Measuring M
M = 0 P0 + 1 P1 ,

|0i

Px = |x i hx | ,
U M = P0 P1

pr(x) = | hx |i |2
|i

UM

|x i

17 / 29

Teleportation
Alice

h |i

|i

Bob

|i = a |0i + b |1i ,

|i

|i =

1 (|00i
2

+ |11i)

18 / 29

Teleportation
Alice

h |i

|i

Bob

|i

|0 i

|i = a |0i + b |1i ,
|0 i =

|i =

1 (|00i
2

+ |11i)

[a |0i (|00i + |11i) + b |1i (|00i + |11i)]

18 / 29

Teleportation
Alice

h |i

|i

Bob

|i

|0 i

|i = a |0i + b |1i ,

|1 i
|i =

1 (|00i
2

+ |11i)

|0 i =

[a |0i (|00i + |11i) + b |1i (|00i + |11i)]

|1 i =

[a |0i (|00i + |11i) + b |1i (|10i + |01i)]

18 / 29

Teleportation
Alice

h |i

|i

Bob

|i

|0 i

|i = a |0i + b |1i ,

|1 i
|i =

|2 i
1 (|00i
2

+ |11i)

|0 i =

[a |0i (|00i + |11i) + b |1i (|00i + |11i)]

|1 i =

[a |0i (|00i + |11i) + b |1i (|10i + |01i)]

|2 i =

1
2
1
2

[a(|0i + |1i)(|00i + |11i) + b(|0i |1i)(|10i + |01i)]

[|00i (a |0i + b |1i) + |01i (a |1i + b |0i)

+ |10i (a |0i b |1i) + |11i (a |1i b |0i)]

18 / 29

Teleportation
Alice

h |i

|i

Bob

|i

|0 i

|i = a |0i + b |1i ,

|1 i
|i =

|2 i
1 (|00i
2

|i
+ |11i)

|0 i =

[a |0i (|00i + |11i) + b |1i (|00i + |11i)]

|1 i =

[a |0i (|00i + |11i) + b |1i (|10i + |01i)]

|2 i =

1
2
1
2

[a(|0i + |1i)(|00i + |11i) + b(|0i |1i)(|10i + |01i)]

[|00i (a |0i + b |1i) + |01i (a |1i + b |0i)

+ |10i (a |0i b |1i) + |11i (a |1i b |0i)]

00 7 |i = a |0i + b |1i ,
10 7 |i = a |0i b |1i ,

01 7 |i = a |1i + b |0i
11 7 |i = a |1i b |0i

18 / 29

Teleportation
Deferred measurement

Alice

|i

|i

Bob

|i

|0 i

|1 i

|2 i

|0 i =

[a |0i (|00i + |11i) + b |1i (|00i + |11i)]

|1 i =

[a |0i (|00i + |11i) + b |1i (|10i + |01i)]

|2 i =

1
2
1
2

[a(|0i + |1i)(|00i + |11i) + b(|0i |1i)(|10i + |01i)]

[|00i (a |0i + b |1i) + |01i (a |1i + b |0i)

+ |10i (a |0i b |1i) + |11i (a |1i b |0i)]

19 / 29

Teleportation
Deferred measurement

Alice

|i

|i

Bob

|i

|0 i

|1 i

|2 i

|3 i

|0 i =

[a |0i (|00i + |11i) + b |1i (|00i + |11i)]

|1 i =

[a |0i (|00i + |11i) + b |1i (|10i + |01i)]

|2 i =

[a(|0i + |1i)(|00i + |11i) + b(|0i |1i)(|10i + |01i)]

1
2
1
2

|3 i =

1
2

(|00i + |01i + |10i + |11i) (a |0i + b |1i)

[|00i (a |0i + b |1i) + |01i (a |1i + b |0i)

+ |10i (a |0i b |1i) + |11i (a |1i b |0i)]

19 / 29

Programmable quantum computer?

Selection
qubits

Working
qubits

n
n

|U i

|i

.
.
.

.
.
.

|P (U, )i

U |i

20 / 29

Programmable quantum computer?

Selection
qubits

Working
qubits

n
n

|U i

.
.
.

|i

.
.
.

|P (U, )i

U |i

1. P does not depend on .

h| i = hU |U i h| i = hP (U, )|P (U, )i h| U U | i
Then hP (U, )|P (U, )i = 1 which implies |P (U, )i = |P (U, )i

20 / 29

Programmable quantum computer?

Selection
qubits

Working
qubits

n
n

|U i

.
.
.

|i

.
.
.

|P (U, )i

U |i

1. P does not depend on .

h| i = hU |U i h| i = hP (U, )|P (U, )i h| U U | i
Then hP (U, )|P (U, )i = 1 which implies |P (U, )i = |P (U, )i
2. If U 6= ei U then hU |U i = 0

hU |U i = hU |U i h|i = hP (U )|P (U )i h| U U |i

Then, either U U = ei I or hP (U )|P (U )i = hU |U i = 0

Already, for one working qubit we have infinitely many different unitary
operators, i.e. we need inifinitely many selection qubits.

20 / 29

Stochastic programmable quantum computer

Selection
qubits
Working
qubit

|U i

x
H

y
U X x Z y |i

|i

1
U = (U I)(|00i + |11i)
2
With a probability 1/4, when x = 0 and y = 0, the output is U |i.
Reading the registers we know when we get the desired result.
For m working qubits 2m selection qubits. Linear dependence!!.
But the probability of success is 2m .

21 / 29

Universal quantum gates

Any unitary operator can be written as the product of others which
are the identity on all but two vectors of the basis
U = Or O2 O1

22 / 29

Universal quantum gates

Any unitary operator can be written as the product of others which
are the identity on all but two vectors of the basis
U = Or O2 O1
Take one such O = W I, s.t. W acts on < |1011i , |0110i >
O

~
W

acts on < |1110i , |0110i >

W

i.e. it is a single qubit gate.

22 / 29

Universal quantum gates

Any unitary operator can be written as the product of others which
are the identity on all but two vectors of the basis
U = Or O2 O1
Take one such O = W I, s.t. W acts on < |1011i , |0110i >
O

~
W

acts on < |1110i , |0110i >

W

i.e. it is a single qubit gate.

And using

we get that CNOT + one qubit gates form a universal set.

DiVicenzo; Sleator & Weinfurter; Barenco et al. (1995)

22 / 29

Physical realization.

Cirac & Zoller (1995)

Trapped ions.

Two energy levels in every ion coupled to a colective phonon.

23 / 29

Physical realization.

Cirac & Zoller (1995)

Trapped ions.

Two energy levels in every ion coupled to a colective phonon.

Single qubit operators
Apply an electromagnetic field of
frequency 0 and phase . The
interaction Hamiltonian is
HI =

~
(X cos + Y sin ) I
2

By controlling the phase and time any U I can be simulated.

23 / 29

Physical realization.

Cirac & Zoller (1995)

SWAP (ion-phonon) gate.

With an electromagnetic field of
frequency 0 Z :
|01i |10i

24 / 29

Physical realization.

Cirac & Zoller (1995)

SWAP (ion-phonon) gate.

With an electromagnetic field of
frequency 0 Z :
|01i |10i
c-Z (ion-phonon) gate.
With a frequency aux + Z :
|11i |11i

24 / 29

Physical realization.

Cirac & Zoller (1995)

SWAP (ion-phonon) gate.

With an electromagnetic field of
frequency 0 Z :
|01i |10i
c-Z (ion-phonon) gate.
With a frequency aux + Z :
|11i |11i
CNOT (ion-ion) gate.
ion-j

ion-j

ion-k

phonon

=
H

ion-k

24 / 29

The rush starts.

Papers per year related to Quantum Information.
14000
12000
10000
8000
6000
4000
2000

1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015

Source, Google Scholar.

Since 1995, the number of papers on Quantum Information has

been growing consistently fast.

25 / 29

Quantum computing

Science 351, 1068-1070 (2016).

26 / 29

Simulation of quantum systems

Science 336, 1130-1133 (2012)

JHEP 04 (2015) 163

Nature Communications 6, 7654 (2015)

27 / 29

Physical realizations.

Science 339 (2013). Special issue

G. Arom, D. Aguil`
a, P. G
amez, F. Luis, O. Roubeau, Chem. Soc. Rev. 41, (2012).

28 / 29

MUCHAS
GRACIAS

29 / 29

MUCHAS
GRACIAS
FELICES VACACIONES

29 / 29