25.12.2014 Views

Introduction to Quantum Computing

Introduction to Quantum Computing

Introduction to Quantum Computing

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Introduction</strong> <strong>to</strong> <strong>Quantum</strong><br />

<strong>Computing</strong><br />

Lecture 1<br />

1


OUTLINE<br />

• Why <strong>Quantum</strong> <strong>Computing</strong><br />

• What is <strong>Quantum</strong> <strong>Computing</strong><br />

• His<strong>to</strong>ry<br />

• <strong>Quantum</strong> Weirdness<br />

• <strong>Quantum</strong> Properties<br />

• <strong>Quantum</strong> Computation<br />

2


Why <strong>Quantum</strong> <strong>Computing</strong><br />

3


Transis<strong>to</strong>rs per chip<br />

10 9<br />

10 8<br />

Transis<strong>to</strong>r Density<br />

<br />

10 7<br />

Pentium<br />

Pro<br />

80786<br />

10 6<br />

10 5<br />

8086<br />

80286<br />

80386<br />

80486<br />

Pentium<br />

10 4<br />

4004<br />

8080<br />

10 3<br />

1970 1975 1980 1985 1990 1995 2000 2005 2010<br />

Year<br />

4


Transis<strong>to</strong>r Size<br />

Electrons per device<br />

10 4<br />

(4M)<br />

10 3<br />

10 2<br />

(16M)<br />

(64M)<br />

(256M)<br />

(1G)<br />

(4G)<br />

(Transis<strong>to</strong>rs per chip)<br />

(16G)<br />

10 1<br />

10 0<br />

10 -1<br />

<br />

1 electron/transis<strong>to</strong>r<br />

1985 1990 1995 2000 2005 2010 2015 2020<br />

Year<br />

5


Why <strong>Quantum</strong> <strong>Computing</strong><br />

• By 2020 we will hit natural limits on the size<br />

of transis<strong>to</strong>rs<br />

• Max out on the number of transis<strong>to</strong>rs per chip<br />

• Reach the minimum size for transis<strong>to</strong>rs<br />

• Reach the limit of speed for devices<br />

• Eventually, all computing will be done using<br />

some sort of alternative structure<br />

• DNA<br />

• Cellular Au<strong>to</strong>ma<strong>to</strong>n<br />

• <strong>Quantum</strong><br />

6


What is <strong>Quantum</strong> <strong>Computing</strong><br />

7


<strong>Introduction</strong><br />

• The common characteristic of any digital<br />

computer is that it s<strong>to</strong>res bits<br />

• Bits represent the state of some physical system<br />

• Electronic computers use voltage levels <strong>to</strong> represent<br />

bits<br />

• <strong>Quantum</strong> systems possess properties that allow<br />

the encoding of bits as physical states<br />

• Direction of spin of an electron<br />

• The direction of polarization of a pho<strong>to</strong>n<br />

• The energy level of an excited a<strong>to</strong>m<br />

8


Spin States<br />

• An electron is always in one of two spin states<br />

• “spin up” – the spin is parallel <strong>to</strong> the particle axis<br />

• “spin down” – the spin is antiparallel <strong>to</strong> the particle<br />

axis<br />

• Notation:<br />

Spin up:<br />

Spin down:<br />

9


qubit<br />

• A qubit is a bit represented by a<br />

quantum system<br />

• By convention:<br />

• A qubit state 0 is the spin up state<br />

• A qubit state 1 is the spin down state<br />

0<br />

1<br />

10


Definitions<br />

• A qubit is governed by the laws of<br />

quantum physics<br />

• While a quantum system can be in one of<br />

a discrete set of states, it can also be in a<br />

blend of states called a superposition<br />

• That is a qubit can be in:<br />

0<br />

1<br />

c 0 0 + c 1<br />

1<br />

|c 0 | 2 +|c 1 | 2 = 1<br />

11


Measurement<br />

• If a qubit is realized by the spin of an<br />

electron, it is possible <strong>to</strong> measure the<br />

qubit value by passing the electron<br />

through a magnetic field<br />

• If the qubit encodes a |0> then it will be<br />

deflected upward<br />

• If the qubit encodes a |1> then it will be<br />

deflected downward<br />

12


Superposition Measurement<br />

• If the qubit is in a superposition state it<br />

cannot be determine if it will deflect up or<br />

down<br />

• However, the probability of each possible<br />

deflection can be found<br />

2<br />

Probability of 0 c 0<br />

2<br />

Probability of 1 c 1<br />

c 0 + c 1<br />

0<br />

1<br />

13


<strong>Quantum</strong> <strong>Computing</strong> His<strong>to</strong>ry<br />

14


His<strong>to</strong>ry<br />

• In the 1970’s s Fredkin, Toffoli, Bennett and others<br />

began <strong>to</strong> look in<strong>to</strong> the possibility of reversible<br />

computation <strong>to</strong> avoid power loss.<br />

• Since quantum mechanics is reversible, a possible link<br />

between computing and quantum devices was suggested<br />

• Some early work on quantum computation occurred<br />

in the 80’s<br />

• Benioff 1980,1982 explored a connection between quantum<br />

systems and a Turing machine<br />

• Feynman 1982, 1986 suggested that quantum systems could<br />

simulate reversible digital circuits<br />

• Deutsch 1985 defined a quantum level XOR mechanism<br />

15


Existing <strong>Quantum</strong> Computers<br />

• liquid NMR quantum computers with 2 –<br />

12 qubit registers.<br />

• Ion Trap method have achieved a single<br />

CONTROLLED NOT and 4 qubit entangled<br />

states<br />

• linear optics,<br />

• Superconductive Device…<br />

16


<strong>Quantum</strong> Weirdness<br />

17


Weird Measurement<br />

• One of the unusual features of<br />

<strong>Quantum</strong> Mechanics is the interaction<br />

between an event and its<br />

measurement<br />

• Measurement changes the state of a<br />

quantum system<br />

• Measurement of the superposition state<br />

of a qubit forces it in<strong>to</strong> one of the qubit<br />

states in an unpredictable manner<br />

18


Comparison I<br />

• Compare qubits <strong>to</strong> classical bits<br />

Assumption Classical <strong>Quantum</strong><br />

A bit always has a<br />

definite value<br />

True<br />

False, a qubit need not have a<br />

definite value until the moment<br />

after it is observed<br />

A bit can only be 0 or 1 True False, a qubit can be in a<br />

superposition of 0 and 1<br />

simultaneously<br />

A bit can be copied without<br />

affecting its value<br />

A bit can be read without<br />

affecting its value<br />

True<br />

True<br />

False, a qubit in an unknown<br />

state cannot be copied without<br />

disrupting its state<br />

False, reading a qubit that is<br />

initially in a superposition will<br />

change the value of the qubit<br />

19


Comparison II<br />

Assumption Classical <strong>Quantum</strong><br />

Reading one bit has no effect<br />

on another unread bit<br />

True<br />

False, if the qubit being read is<br />

entangled with another qubit<br />

reading one will affect the other<br />

20


<strong>Quantum</strong> Phenomena<br />

21


<strong>Quantum</strong> Phenomena<br />

• There are five quantum phenomena<br />

that make quantum computing weird<br />

• Superposition<br />

• Interference<br />

• Entanglement<br />

• Non-determinism<br />

• Non-clonability<br />

22


Superposition<br />

• The Principal of Superposition states if a<br />

quantum system can be measured <strong>to</strong> be in<br />

one of a number of states then it can also<br />

exist in a blend of all its states<br />

simultaneously<br />

• RESULT: An n-bit n<br />

qubit register can be in all<br />

2 n states at once<br />

• Massively parallel operations<br />

23


Interference<br />

• We see interference patterns when light<br />

shines through multiple slits<br />

• This is a quantum<br />

phenomena which is<br />

also present in quantum<br />

computers<br />

• A quantum computer<br />

can operate on several<br />

inputs at once, the results<br />

interfere with each other<br />

producing a collective<br />

result<br />

24


Entanglement<br />

• If two or more qubits are made <strong>to</strong> interact,<br />

they can emerge from the interaction in a joint<br />

quantum state which is different from any<br />

combination of the individual quantum states<br />

• RESULT: If two entangled qubits are<br />

separated by any distance and one of them is<br />

measured then the other, at the same instant,<br />

enters a predictable state<br />

25


Non-Determinism<br />

• <strong>Quantum</strong> non-determinism refers <strong>to</strong> the<br />

condition of unpredictability<br />

• If a quantum system is in a superposition<br />

state and then measured, the measured<br />

state can not be predicted.<br />

26


Non-Clonability<br />

• It is impossible <strong>to</strong> copy an unknown<br />

quantum state exactly<br />

• If you asked a friend <strong>to</strong> prepare a qubit in a<br />

superposition state without telling you<br />

which superposition state, then you could<br />

not make a perfect copy of the qubit<br />

• Useful in quantum cryp<strong>to</strong>logy<br />

27


<strong>Quantum</strong> Computation<br />

28


<strong>Quantum</strong> Computation<br />

Changes <strong>to</strong> a quantum state can be described using the<br />

language of quantum computation<br />

• Single Qubit Gates<br />

Classical Not Gate<br />

- Truth table<br />

0 →1 and 1→0<br />

<strong>Quantum</strong> Not Gate - Truth table<br />

0 → 1 and 1 → 0<br />

29


<strong>Quantum</strong> Computation<br />

Superposition of states<br />

Not without further knowledge of the properties of<br />

quantum gates<br />

The quantum NOT gate acts LINEARLY…<br />

α 0 + β 1 → α 1 + β 0<br />

Linear behaviour is a general property of quantum<br />

mechanics<br />

Non-linear<br />

behaviour can lead <strong>to</strong> apparent paradoxes<br />

- Time Travel<br />

- Faster than light communication<br />

- Violates the 2 nd Law of Thermodynamics<br />

30


<strong>Quantum</strong> Computation<br />

NOT gate representation<br />

X<br />

we get…<br />

⎡0 1⎤<br />

≡ ⎢<br />

1 0 ⎥<br />

⎣ ⎦<br />

for any<br />

⎡α<br />

⎤<br />

α 0 + β 1 ≡ ⎢<br />

β ⎥<br />

⎣ ⎦<br />

⎡α⎤ ⎡0 1⎤⎡α⎤ ⎡β⎤<br />

X ⎢ or β 0 α 1<br />

β<br />

⎥ = ⎢ = +<br />

1 0<br />

⎥⎢<br />

β<br />

⎥ ⎢<br />

α<br />

⎥<br />

⎣ ⎦ ⎣ ⎦⎣ ⎦ ⎣ ⎦<br />

<strong>to</strong> summarize…<br />

α 0 + β 1 → α 1 + β 0<br />

31


<strong>Quantum</strong> Computation<br />

Are there any constraints on what matrices may be used as<br />

quantum gates Of course!<br />

We require the normalization condition<br />

α<br />

2 2<br />

+ β = 1 ψ = α 0 + β 1<br />

for<br />

and the result ψ ' = α' 0 + β'1<br />

after the gate has<br />

acted<br />

The appropriate condition for this (of course) is<br />

that the matrix representing the gate is<br />

UNITARY<br />

†<br />

UU=<br />

I<br />

†<br />

where U is the adjoint of<br />

That's it!!! Anything else is a valid quantum gate.<br />

U<br />

32


<strong>Quantum</strong> Computation<br />

Two more important gates…<br />

• Z gate<br />

Z<br />

⎡1 0⎤<br />

≡ ⎢<br />

0 -1 ⎥<br />

⎣ ⎦<br />

leaves 0 unchanged<br />

flips the sign of 1 <strong>to</strong> - 1<br />

• Hadamard Gate<br />

H<br />

1 ⎡1 1⎤<br />

≡ ⎢<br />

2 1 -1<br />

⎥<br />

⎣ ⎦<br />

( + )<br />

( − )<br />

turns 0 in<strong>to</strong> 0 1 2<br />

turns 1 in<strong>to</strong> 0 1 2<br />

Note: Applying H twice <strong>to</strong> a state does nothing <strong>to</strong> it.<br />

H<br />

2<br />

=<br />

I<br />

33


<strong>Quantum</strong> Computation<br />

Hadamard Gate: A most useful gate indeed!<br />

if H =<br />

1<br />

( X + Z)<br />

and ψ = α 0 + β 1 then<br />

2<br />

H ψ =<br />

1<br />

( X ψ + Z ψ )<br />

2<br />

=<br />

1 ⎛⎡0 1⎤⎡α ⎤ ⎡1 0 ⎤⎡α⎤⎞ 1 ⎛⎡β⎤ ⎡α ⎤⎞ 1 ⎡α + β⎤<br />

⎜⎢ 2 1 0<br />

⎥⎢<br />

β<br />

⎥+ ⎢ ⎟= ⎜ + ⎟=<br />

0 −1<br />

⎥⎢<br />

β<br />

⎥ ⎢<br />

2 α<br />

⎥ ⎢<br />

−β ⎥ ⎢<br />

2 α −β<br />

⎥<br />

⎝⎣ ⎦⎣ ⎦ ⎣ ⎦⎣ ⎦⎠ ⎝⎣ ⎦ ⎣ ⎦⎠<br />

⎣ ⎦<br />

for<br />

for<br />

H<br />

H<br />

1<br />

0 α=1, β=0 H 0 = ( 0 + 1 )<br />

2<br />

1<br />

1 α = 0, β = 1 H 1 = 0 − 1<br />

2<br />

( )<br />

34


<strong>Quantum</strong> Computation<br />

• Review: Important single-qubit<br />

gates<br />

α 0 + β 1<br />

α 0 + β 1<br />

α 0 + β 1<br />

X<br />

Z<br />

H<br />

β 0 + α 1<br />

α 0 − β 1<br />

0 + 1 1 − 1<br />

α + β<br />

2 2<br />

35


<strong>Quantum</strong> Computation<br />

• Arbitrary Single Qubit <strong>Quantum</strong> Gate<br />

- complete set from properties of a much smaller set<br />

U<br />

β<br />

δ<br />

⎡<br />

γ γ<br />

−i<br />

⎤⎡<br />

⎤<br />

2<br />

cos sin ⎡ −i<br />

⎤<br />

2<br />

i<br />

e 0 ⎢<br />

−<br />

2 2⎥<br />

α<br />

e 0<br />

= e<br />

⎢ ⎥ ⎢ ⎥<br />

⎢ β ⎥⎢ ⎥<br />

δ<br />

i γ γ ⎢ ⎥<br />

i<br />

⎢ 2 ⎢sin<br />

cos ⎥<br />

2<br />

⎣ 0 e ⎥⎦ ⎢ 0 e ⎥<br />

⎢⎣<br />

2 2 ⎥⎦⎣ ⎦<br />

Global<br />

Phase<br />

Fac<strong>to</strong>r<br />

Rotation<br />

about z<br />

Rotation<br />

Scaling<br />

Constant<br />

α, βγ , and δare<br />

all real valued<br />

36


<strong>Quantum</strong> Computation<br />

• Classical Universal Gates (example)<br />

- The NAND gate is a classical Universal Gate. Why<br />

NOT gate using NAND AND gate using NAND OR gate using NAND<br />

• Universal <strong>Quantum</strong> Gates<br />

- An arbitrary quantum Computation on n qubits can be<br />

generated by a finite set of gates that are UNIVERSAL<br />

for quantum computation<br />

* Need <strong>to</strong> introduce some multiple quibit quantum gates<br />

37


Multiple Qubit Gates<br />

• Controlled-NOT (CNOT) Gate<br />

- two input qubits: : control and target<br />

A<br />

A<br />

B<br />

B<br />

⊕<br />

A<br />

if control is 0 target left alone 00 → 00 or 01 → 01<br />

else control is 1 target qubit is flipped 10 → 11 or 11 → 10<br />

- In General<br />

AB , → AB , ⊕A<br />

38


CNOT quantum gate<br />

⎡1<br />

0 0 0⎤<br />

A<br />

A<br />

⎢<br />

0 1 0 0<br />

⎥<br />

U CN<br />

= ⎢ ⎥<br />

⎢ 0 0 0 1 ⎥<br />

B<br />

B ⊕ A<br />

⎢ ⎥<br />

⎣0<br />

0 1 0⎦<br />

⎡B0<br />

⎤<br />

A0 = 1<br />

⎢<br />

B<br />

⎥<br />

1<br />

if A = 0 then we get ⎢ ⎥<br />

⎡ ⎡B0⎤⎤ ⎡AB<br />

A<br />

0 0⎤<br />

1<br />

= 0 ⎢ 0 ⎥<br />

⎢A0<br />

⎢ B<br />

⎥⎥<br />

⎢ ⎢ ⎥<br />

1 AB<br />

⎥<br />

⎣ ⎦<br />

0<br />

0 1<br />

⎣ ⎦<br />

A B →<br />

⎢ ⎥<br />

= ⎢ ⎥<br />

⎢ ⎡B<br />

⎤⎥ ⎢AB<br />

⎥ ⎡ 0 ⎤<br />

0<br />

1 0<br />

⎢A1<br />

⎢<br />

B<br />

⎥⎥ AB<br />

1<br />

⎢ 1 1⎥<br />

A0<br />

= 0<br />

⎢<br />

0<br />

⎥<br />

⎢ ⎥ ⎣ ⎦<br />

⎣ ⎣ ⎦⎦ if A = 1 then we get ⎢ ⎥<br />

A1 = 1 ⎢B<br />

⎥<br />

0<br />

⎢⎣<br />

B1<br />

⎥⎦<br />

Any multiple qubit<br />

qubit logic gate may be composed from<br />

CNOT and Single Qubit Gates<br />

39


Other Computational Bases<br />

• Measurements<br />

( 0 + 1 ) ( 0 − 1 )<br />

- In terms of + = , − =<br />

basis states<br />

2 2<br />

+ + − + − − α + β α −β<br />

ψ = α 0 + β 1 = α + β = + + −<br />

2 2 2 2<br />

- Generally any basis state can represent an arbitrary<br />

qubit state<br />

ψ = α a + β b<br />

- If orthonormal then we can perform a measurement in<br />

keeping with probability interpretation<br />

40


<strong>Quantum</strong> Circuits<br />

• Elements of a <strong>Quantum</strong> Circuit<br />

- each line in a circuit represents a "wire"<br />

* passage of time<br />

* pho<strong>to</strong>n moving from one location <strong>to</strong> another<br />

- assume the state input is a computational basis state<br />

- input is usually the state consisting of all 0 s<br />

- no loops allowed ie: : acyclic<br />

- No FANIN(not reversible therefore not Unitary)<br />

- FANOUT (can't copy a qubit)<br />

41


<strong>Quantum</strong> Circuits<br />

• <strong>Quantum</strong> Qubit Swap Circuit<br />

ab , → aa , ⊕b<br />

( )<br />

( )<br />

→ a⊕ a⊕b , a⊕ b = b,<br />

a⊕b<br />

→ b, a⊕b ⊕ b = b,<br />

a<br />

a<br />

b<br />

aa ,<br />

⊕ b<br />

ba ,<br />

⊕ b<br />

ba ,<br />

x<br />

x<br />

42


• Controlled-U U Gate<br />

<strong>Quantum</strong> Circuits<br />

- A Controlled-U U Gate has one control qubit and n target<br />

qubits<br />

- where U is any unitary matrix acting on n qubits<br />

U<br />

43


<strong>Quantum</strong> Circuits<br />

• Measurement Operation<br />

- Converts a single qubit state in<strong>to</strong> a probabilistic<br />

classical bit M<br />

ψ<br />

M<br />

44


<strong>Quantum</strong> Circuits<br />

• Can we make a Qubit Copying Circuit<br />

- Copying a classical bit can be done with the<br />

Classical CNOT gate<br />

bit <strong>to</strong> be<br />

copied<br />

x<br />

x<br />

x<br />

x<br />

original<br />

bit<br />

0<br />

y x⊕<br />

y<br />

x<br />

scratch-pad<br />

initialized <strong>to</strong> zero<br />

copied<br />

bit<br />

45


<strong>Quantum</strong> Circuits<br />

• Can we make a Qubit Copying Circuit<br />

- How about copying a qubit in an unknown state using<br />

a controlled-CNOT gate<br />

ψ = a 0 + b 1<br />

bit <strong>to</strong> be<br />

copied<br />

a<br />

0 + b 1<br />

Output State<br />

a<br />

00 + b 10<br />

0<br />

a<br />

00 + b 11<br />

scratch-pad<br />

initialized <strong>to</strong> zero<br />

46


<strong>Quantum</strong> Circuits<br />

• Can we make a Qubit Copying Circuit<br />

- Does ψ ψ = a 00 + b 11<br />

( )( )<br />

2 2<br />

ψ ψ = a 0 + b 1 a 0 + b 1 = a 00 + ab 01 + ab 10 + b 11<br />

- Unless ab = 0this does not copy the quantum state<br />

input<br />

2 2<br />

a 00 + ab 01 + ab 10 + b 11 ≠ a 00 + b 11<br />

- It is impossible <strong>to</strong> make a copy of the unknown<br />

quantum state<br />

- NO CLONING THEOREM -<br />

47


<strong>Quantum</strong> Circuits<br />

• Bell States, EPR States, EPR Pairs<br />

x<br />

y<br />

H<br />

Out<br />

00<br />

β xy 01<br />

10<br />

11<br />

+ ⎫<br />

⎪<br />

2 ⎬ → →<br />

→ ⎪ ⎭<br />

0 1<br />

0 → 00 + 10 00 + 11<br />

0<br />

In<br />

Out<br />

( 00 + 11 ) 2 ≡ β00<br />

( 01 + 10 ) 2 ≡ β01<br />

( 00 + 11 ) 2 ≡ β00<br />

( 00 + 11 ) 2 ≡ β00<br />

2 2<br />

48


<strong>Quantum</strong> Algorithms<br />

Initial State<br />

( )<br />

xy , → xy , ⊕ f x<br />

Final State<br />

Data Register<br />

0 + 1<br />

2<br />

0<br />

x<br />

U f<br />

x<br />

y y⊕<br />

f ( x)<br />

ψ<br />

Target Register<br />

00 + 10<br />

xy , =<br />

2<br />

ψ<br />

0,0 ⊕ f 0 + 1,0 ⊕ f 1 0, f 0 + 1, f 1<br />

= =<br />

2 2<br />

( ) ( ) ( ) ( )<br />

49


<strong>Quantum</strong> Algorithms<br />

Eureka!!!! Both values of the function<br />

show up in the final state solution.<br />

ψ<br />

=<br />

0, f 0 + 1, f 1<br />

( ) ( )<br />

2<br />

This can be generalized <strong>to</strong> functions on<br />

arbitrary number of bits using the…<br />

HADAMARD TRANSFORM<br />

or<br />

WALSH-HADAMARD HADAMARD TRANSFORM<br />

50


<strong>Quantum</strong> Algorithms<br />

• Deutsch's Algorithm Circuit<br />

- Combines quantum parallelism and interference<br />

0 H<br />

0 + 1<br />

2<br />

x<br />

x<br />

H<br />

1 H<br />

0 − 1<br />

2<br />

U f<br />

y y⊕<br />

f ( x)<br />

↑<br />

↑<br />

↑<br />

↑<br />

ψ 0<br />

ψ 1<br />

ψ 2<br />

ψ 3<br />

51


<strong>Quantum</strong> Algorithms<br />

• Deutsch's Algorithm Calculations<br />

- Combines quantum parallelism and interference<br />

ψ<br />

0<br />

= 01<br />

ψ<br />

ψ<br />

⎡ 0 + 1 ⎤⎡ 0 − 1 ⎤<br />

→ ψ =⎢ ⎥⎢ ⎥<br />

⎣ 2 ⎦⎣ 2 ⎦<br />

0 1<br />

⎧ ⎡ 0 + 1 ⎤⎡ 0 − 1 ⎤<br />

⎪± ⎢ ⎥⎢ ⎥ if f 0 = f 1<br />

⎪ ⎣ 2 ⎦⎣ 2 ⎦<br />

→ ψ =⎨<br />

⎪ ⎡ 0 − 1 ⎤⎡ 0 − 1 ⎤<br />

⎪ ± ⎢ ⎥⎢ ⎥ if f 0 ≠ f 1<br />

⎩ ⎣ 2 ⎦⎣ 2 ⎦<br />

1 2<br />

( ) ( )<br />

( ) ( )<br />

52


<strong>Quantum</strong> Algorithms<br />

• Deutsch's Algorithm Conclusion<br />

ψ<br />

⎧ ⎡ 0 − 1 ⎤<br />

⎪± 0 ⎢ ⎥ if f 0 = f 1<br />

⎪ ⎣ 2 ⎦<br />

→ ψ =⎨<br />

⎪ ⎡ 0 − 1 ⎤<br />

⎪ ± 1 ⎢ ⎥ if f 0 ≠ f 1<br />

⎩ ⎣ 2 ⎦<br />

2 3<br />

( ) ( )<br />

( ) ( )<br />

realizing f ( 0) ⊕ f ( 1)<br />

is 0 if f ( 0) = f ( 1)<br />

and 1 otherwise…<br />

ψ<br />

3<br />

f<br />

0 f 1<br />

⎡ 0 − 1 ⎤<br />

=± ( ) ⊕ ( ) ⎢ ⎥<br />

⎣<br />

2<br />

⎦<br />

measuring the 1 st qubit gives f ( 0) ⊕ f ( 1)<br />

53


<strong>Quantum</strong> Algorithms<br />

• Deutsch's Algorithm Results<br />

- The quantum circuit has given us the ability <strong>to</strong><br />

determine a GLOBAL PROPERTY of f ( x)<br />

namely<br />

f ( 0) ⊕ f ( 1)<br />

using only ONE evaluation of<br />

f ( x)<br />

- A classical computer would require at least two<br />

evaluations!<br />

- Difference between quantum parallelism and classical<br />

randomized algorithms<br />

( ) ( )<br />

* One might think the state 0 f 0 + 1 f 1 corresponds <strong>to</strong><br />

probabilistic classical computer that evaluates f ( 0)<br />

with probability 1/2<br />

or f () 1 with probability ½. These are classically mutually exclusive.<br />

* <strong>Quantum</strong> mechanically these two alternatives can INTERFERE <strong>to</strong><br />

yield some global property of the function f and by using a Hadamard gate<br />

can recombine the different alternatives<br />

54


<strong>Quantum</strong> Algorithms<br />

• Deutsch-Jozsa<br />

Algorithm<br />

- A simple case of a more general algorithm<br />

- Application is called Deutsch's Problem<br />

x is a number<br />

from 0 <strong>to</strong> 2 n -1<br />

Alice<br />

x<br />

n bits each time<br />

Bob<br />

f<br />

( x)<br />

⎧ Constant for all values of x<br />

⎨<br />

⎩Balanced: 1 for 1/ 2 the values of x or 0 otherwise<br />

- Classically Alice can only send one value of x each<br />

time<br />

- Best classical algorithm requires up <strong>to</strong> 2 n /2+<br />

1queries<br />

n<br />

2 /2 0 's and one 1⇒<br />

Balanced<br />

55


ψ 0<br />

ψ 1<br />

ψ 2<br />

ψ 3<br />

<strong>Quantum</strong> Algorithms<br />

• Deutsch-Jozsa<br />

Algorithm<br />

- If Bob and Alice were able <strong>to</strong> exchange qubits instead<br />

of classical bits and if Bob calculated f(x) using a unitary<br />

transform U f then Alice could determine the function in<br />

one query.<br />

- Alice has an n qubit register and a single qubit register<br />

which she gives <strong>to</strong> Bob<br />

- Prepares query and answer register in a superposition<br />

state<br />

- Bob evaluates f(x) ) and puts result in<strong>to</strong> answer register<br />

- Alice interferes the states in the superposition using a<br />

hadamard transform on the query register<br />

56


<strong>Quantum</strong> Algorithms<br />

• Deutsch-Jozsa<br />

Algorithm Circuit<br />

0<br />

n<br />

⊗n<br />

H<br />

x<br />

x<br />

⊗n<br />

H<br />

U f<br />

1 H<br />

↑<br />

↑<br />

y y⊕<br />

f ( x)<br />

↑<br />

↑<br />

ψ 0<br />

ψ 1<br />

ψ 2<br />

ψ 3<br />

ψ<br />

⊗n<br />

x ⎡ 0 + 1 ⎤<br />

= 0 1 → ψ = ∑ ⎢ ⎥<br />

n<br />

n<br />

x∈<br />

2 ⎣ 2 ⎦<br />

0 1<br />

{ 0,1}<br />

ψ<br />

1 2<br />

{ 0,1}<br />

f<br />

( ) ( x )<br />

−1 x ⎡ 0 − 1 ⎤<br />

→ ψ = ∑<br />

⎢ ⎥<br />

n<br />

n<br />

x∈<br />

2 ⎣ 2 ⎦<br />

Bob's function<br />

evaluation is<br />

s<strong>to</strong>red in the<br />

amplitude<br />

57


Hadamard transform: helps <strong>to</strong> calculate effect on a state x<br />

By checking the cases x=0 and x=1 separately for a single qubit…<br />

thus<br />

<strong>Quantum</strong> Algorithms<br />

• Deutsch-Jozsa<br />

Algorithm - de<strong>to</strong>ur<br />

H x =<br />

∑<br />

( −1)<br />

where is the bitwise inner product of x and z, modulo 2<br />

z<br />

xz<br />

z<br />

2<br />

( ) 1 1<br />

⊗n<br />

H x ,..., x = ∑ −1<br />

x•<br />

z<br />

xz+ ... + x z 1 n<br />

n n<br />

n z1<br />

zn<br />

n<br />

1 ,...,<br />

H<br />

⊗n<br />

x<br />

= ∑ −<br />

z<br />

1<br />

( )<br />

x•<br />

z<br />

z<br />

2<br />

n<br />

z<br />

,...,<br />

2<br />

z<br />

58


<strong>Quantum</strong> Algorithms<br />

• Deutsch-Jozsa<br />

Algorithm Circuit<br />

ψ<br />

2 3<br />

- amplitude for is…<br />

( )<br />

−1 z ⎡ 0 − 1 ⎤<br />

→ ψ = ∑∑<br />

n ⎢ ⎥<br />

z x 2 ⎣ 2 ⎦<br />

0 n<br />

Case 1: If f is constant the amplitude for is +1 or -1<br />

depending on the constant value f(x) ) takes. Since<br />

ψ<br />

is unit length then all other amplitudes must be zero.<br />

- An observation will yield 0s for all qubits in the register<br />

( )<br />

x• z+<br />

f x<br />

query register<br />

f<br />

( ) ( x )<br />

⊗<br />

∑<br />

x<br />

−1<br />

2<br />

n<br />

0 n<br />

⊗ 3<br />

59


<strong>Quantum</strong> Algorithms<br />

• Deutsch-Jozsa<br />

Algorithm Circuit<br />

Case 2: If f is balanced then the positive and negative<br />

contributions <strong>to</strong> the amplitude for 0 ⊗n<br />

cancel, leaving an<br />

amplitude of 0<br />

- A measurement must yield a result other than 0 on at<br />

least one qubit<br />

Summary:<br />

- If Alice measures all zeros then the function is constant<br />

- Otherwise the function is balanced.<br />

- Deutsch's problem on a quantum computer can be<br />

solved in one evaluation.<br />

60


<strong>Quantum</strong> Algorithms<br />

• Other <strong>Quantum</strong> Algorithms<br />

- Generally there are three classes<br />

* Discrete Fourier Transform Algorithms<br />

~Deutsch-Jozsa<br />

Algorithm<br />

~Shor's<br />

Algorithm for Fac<strong>to</strong>ring<br />

~Shor's<br />

Discrete Logarithm Algorithm<br />

* <strong>Quantum</strong> Search Algorithms<br />

* <strong>Quantum</strong> Simulation Algorithms<br />

~<strong>Quantum</strong> Computer is used <strong>to</strong><br />

simulate quantum systems<br />

61


Experimental <strong>Quantum</strong> Information<br />

Processing<br />

• The Stern-Gerlach<br />

Experiment<br />

• Optical Techniques<br />

• Nuclear Magnetic Resonance<br />

• <strong>Quantum</strong> Dots<br />

• Traps: Ion Traps & Neutral A<strong>to</strong>m Traps<br />

62


NMR <strong>Quantum</strong> <strong>Computing</strong><br />

Lecture 2<br />

63


Nuclear Magnetic Resonance<br />

<strong>Quantum</strong> Computers<br />

Qubit representation: spin of an a<strong>to</strong>mic nucleus<br />

Unitary evolution: using magnetic field pulses<br />

applied <strong>to</strong> spins in a strong magnetic<br />

field.<br />

Chemical bonds between a<strong>to</strong>ms couple the spins<br />

State preparation: using a strong magnetic field <strong>to</strong><br />

polarize the spins<br />

Readout: using magnetic-moment induced free<br />

induction decay signals<br />

64


Nuclear Magnetic Resonance Q.C.<br />

Physical Apparatus<br />

pre - amplifier<br />

RF -source<br />

Computer<br />

Liquid sample<br />

12<br />

C,<br />

RF - coil<br />

19<br />

F,<br />

15<br />

N,<br />

31<br />

P<br />

=11.8 Tesla<br />

B (uniform <strong>to</strong><br />

1 part in 10 9 )<br />

Regard as an ensemble<br />

of n-bit quantum<br />

computers<br />

amplifier<br />

Typical Experiment<br />

Wait a few minutes<br />

for the sample <strong>to</strong> come<br />

<strong>to</strong> thermal equilibrium<br />

2. Send RF pulses <strong>to</strong><br />

manipulate nuclear spins<br />

in<strong>to</strong> desired state.<br />

3. Switch off the amps<br />

and switch on the preamplifier<br />

<strong>to</strong> measure<br />

the free-induction decay<br />

65


Nuclear Magnetic Resonance Q.C.<br />

Spectrometer<br />

Physical Apparatus<br />

Nuclear Spins as qubits<br />

ADC for data acquisition<br />

RF synthesizer and amplifier<br />

Gradient control<br />

0<br />

1<br />

B<br />

wave guides<br />

sample<br />

test tube<br />

9.6 T<br />

I<br />

J IS<br />

S<br />

RF Wave<br />

RF wave<br />

High field magnet<br />

2-3 Dibromothiophene<br />

66


Internal Hamil<strong>to</strong>nian<br />

• The evolution of a spin system is<br />

generated by Hamil<strong>to</strong>nians<br />

• Internal Hamil<strong>to</strong>nian:<br />

H int =ω I I z +ω S S z +2π J IS I z S z<br />

interaction with B field<br />

9.6 T<br />

I<br />

J IS<br />

S<br />

spin-spin coupling<br />

2-3 Dibromothiophene<br />

67


External Hamil<strong>to</strong>nian<br />

• Experimentally Controlled Hamil<strong>to</strong>nian:<br />

H ext (t) =ω RFx (t)·(I<br />

(I x +S x )+<br />

spins couple <strong>to</strong> RF field<br />

)+ω RFy (t)<br />

(t)·(I(I y +S y )<br />

• Total Hamil<strong>to</strong>nian:<br />

H <strong>to</strong>tal (t) = H int + H ext (t)<br />

H <strong>to</strong>tal (t<br />

(t)<br />

controlled via<br />

H ext (t)<br />

9.6 T<br />

I J IS S<br />

RF wave<br />

2-3 Dibromothiophene<br />

68


Tomography<br />

Not all elements of the density matrix are observable on an<br />

NMR spectra.<br />

2<br />

σ<br />

x<br />

σ<br />

2 3<br />

x<br />

σ<br />

z<br />

To observe the other elements of the density matrix<br />

requires repeating the experiment 7 times with<br />

readout pulses appended <strong>to</strong> the pulse program.<br />

This is done without changing any other parameters<br />

of the pulse program.<br />

69


One Example of NMR QC:<br />

<strong>Quantum</strong> Games:<br />

theoretical and experimental results<br />

70


Outline<br />

• <strong>Introduction</strong> of quantum games<br />

• Classical game: Prisoner’s s Dilemma<br />

• Maximal entangled quantum game<br />

• Some of our results<br />

• Theoretical extensions with non-maximal<br />

entanglement, more players, larger<br />

strategy space, and so on.<br />

• Experimental realization of quantum game<br />

• Future Plan and discussion<br />

71


• Game theory<br />

Prisoner's dilemma<br />

--an important branch of applied mathematics. It is the<br />

theory of decision-making and conflict between different agents.<br />

Since the seminal book of Von Neumann and Morgenstern,<br />

modern game theory has found applications ranging from<br />

economics through <strong>to</strong> biology.<br />

• It concludes: Players, Strategy space, Payoff function<br />

Classifications: Time (static & Dynamic).<br />

Information (complete &incomplete)<br />

• Prisoner’s s Dilemma<br />

--a a famous game in game theory.<br />

72


• Table: Payoff matrix for the Prisoner's Dilemma. The first entry in the<br />

parenthesis denotes the payoff of Alice and the second <strong>to</strong> Bob's.<br />

Alice<br />

C<br />

D<br />

C<br />

(3,3)<br />

(5,0)<br />

Bob<br />

D<br />

(0,5)<br />

(1,1)<br />

• Nash Equilibrium: mutual defect (D,D)<br />

• Nash Equilibrium implies that no player can increase his payoff by<br />

unilaterally changing his strategy.<br />

• Pare<strong>to</strong> optimal: mutual cooperation (C,C)<br />

• A pair of strategies is called pare<strong>to</strong> optimal if it is not possible <strong>to</strong><br />

increase one player’s s payoff without lessening the payoff of the<br />

other player.<br />

• Prisoner’s s Dilemma: Nash Equilibrium strategy<br />

profile is not equivalent <strong>to</strong> Pare<strong>to</strong> optimal<br />

73


Maximal entangled quantum game<br />

• <strong>Quantum</strong> game theory<br />

Recently, new effect involving quantum information on has<br />

been discovered theoritically in the area of game theory by<br />

some pioneers.<br />

1. L.Goldenberg, L.Vaidman, S.Wiesner, Phys.Rev.Lett. . 82, 3356<br />

(1999).<br />

2. D.A.Meyer, Phys.Rev.Lett. . 82, 1052 (1999).<br />

3. J.Eisert, M.Wilkens, M.Lewenstein, Phys.Rev.Lett. . 83, 3077<br />

(1999).<br />

• Maximal entangled quantum game<br />

Eisert et al. showed that the classical problem of<br />

Prisoner's Dilemma is a subset of the quantum game by<br />

using a physical model of the quantum game, , and there is no<br />

longer a dilemma when employ a maximally entangled game.<br />

74


Interesting results<br />

For a separable game with γ =0 ,<br />

there exists a pair of quantum strategies (D,D) is a<br />

Nash Equilibrium and yields payoff (1,1) which is not Pare<strong>to</strong><br />

optimal. Indeed, this quantum game behaves “classically”.<br />

For a maximally entangled quantum game<br />

π<br />

γ =<br />

2<br />

with ,<br />

(Q,Q) is the Nash equilibrium of the game and has the<br />

property <strong>to</strong> be Pare<strong>to</strong> optimal .<br />

So Prisoner’s s Dilemma is removed<br />

if quantum strategies are allowed for.<br />

75


Correlation between quantum game<br />

and quantum entanglement<br />

• Yet it is legitimate for us <strong>to</strong> ask:<br />

--Whether a quantum game will still<br />

outperform its classical version if it is not<br />

maximally entangled and how a quantum<br />

game depends on the entanglement of the<br />

game's state<br />

76


A physical model of quantum game<br />

• J. Eisert have proposed a physical model of this game<br />

and the elegant quantum network is illustrated as:<br />

• U A and U B are the strategy moves available <strong>to</strong> the<br />

players:<br />

⎛<br />

iφ<br />

⎞<br />

( ) = ⎜<br />

e cos θ sin θ<br />

U θ , φ<br />

2 2 ⎟<br />

• Unitary opera<strong>to</strong>r<br />

J<br />

{ i D ⊗ 2}<br />

⎜<br />

⎝<br />

− sin θ<br />

2<br />

e<br />

−iφ<br />

cos θ ⎟<br />

2 ⎠<br />

⎛ 0<br />

⎜<br />

⎝−1<br />

1⎞<br />

0⎠<br />

= exp γ D/<br />

D=<br />

Uπ ( ,0) = ⎟<br />

^<br />

77


• Two player’s s initial state is<br />

( ) ( )11<br />

ψi = J 00 = cos γ 2 00 + isin<br />

γ 2<br />

• The entanglement of the game's initial state can be<br />

denoted as<br />

γ γ γ<br />

γ<br />

− sin<br />

therefore, can be denoted as a measure for the<br />

entanglement.<br />

• The final state is<br />

ψ = J<br />

+ U ⊗U<br />

J<br />

f<br />

2<br />

ln sin<br />

− cos<br />

2<br />

( ) 00<br />

• Then the expected payoff for Alice and Bob are<br />

A<br />

B<br />

2<br />

ln cos<br />

2 2<br />

2<br />

2 γ<br />

ξ<br />

Α<br />

= 3P00<br />

+ 5P10<br />

+ P11<br />

2<br />

ξ<br />

B<br />

= 3P00<br />

+ 5P01<br />

+ P<br />

Pij<br />

= ij ψ<br />

f<br />

11<br />

2<br />

78


• Nash Equilibrium:<br />

Uˆ<br />

A<br />

⊗Uˆ<br />

B<br />

Theoretical Results<br />

⎧ Dˆ<br />

⊗ Dˆ , 0 ≤ γ ≤ γ<br />

th1<br />

⎪<br />

Dˆ<br />

⊗Qˆ,<br />

γ<br />

th1<br />

≤ γ ≤ γ<br />

th2<br />

= ⎨<br />

,<br />

⎪Q<br />

⊗ Dˆ , γ<br />

th1<br />

≤ γ ≤ γ<br />

th2<br />

⎪<br />

⎩ Qˆ<br />

⊗Qˆ,<br />

γ<br />

th2<br />

≤ γ ≤ π 2<br />

• There exist two threshold:<br />

γ sin ( 1 5 ) γ sin ( 2 5 )<br />

th 1<br />

= Arc<br />

⎛<br />

Dˆ<br />

0<br />

= ⎜<br />

⎝−1<br />

th<br />

=<br />

2<br />

Arc<br />

−1⎞<br />

⎟,<br />

0 ⎠<br />

⎛ i<br />

Qˆ<br />

= ⎜<br />

⎝0<br />

• Expected payoff as game’s s entanglement varies<br />

0 ⎞<br />

⎟<br />

− i⎠<br />

79


Other Theoretical Results<br />

• Differnet sets of strategies.<br />

J. Du et al., Physics Letter A, 289 (2001) 9<br />

• Multi players more than 2-player.<br />

J. Du et al., Physics Letter A, 302 (2002) 229<br />

• Phase-transition-like behavior of quantum games<br />

J. Du et al., Journal of Physics A: Mathematical and General 36,<br />

6551-6562 (2003) .<br />

• One Review<br />

J. Du et al., Fluctuation and Noise Letters Vol 2, Iss 4, R189-R203.<br />

• <strong>Quantum</strong> games in econophysics<br />

H. Li, J. Du and S. Massar, Physics Letter A, 306 (2002) 73<br />

J. Du et al., Physics Review E 68, 016124 (2003)<br />

80


Experimental realization<br />

Physics Review Letter 88, , 137902(2002<br />

2002)<br />

• Technologies for quantum information<br />

processing(QIP)<br />

-There are a number of proposed device technologies for QIP.<br />

--Of them, NMR have given the many successful results<br />

experimentally for QIP, such as quantum teleportation, quantum<br />

error correction, quantum simulation, quantum algorithm etc.<br />

• We add game theory <strong>to</strong> the list: <strong>Quantum</strong><br />

games was experimental realized on nuclear<br />

magnetic resonance quantum computer.<br />

81


• Qubits<br />

Two-qubit<br />

qubit: : Nuclear Coupled Spins<br />

Partially deuterated cy<strong>to</strong>sine<br />

molecule contains two pro<strong>to</strong>ns, in a<br />

magnetic field, each spin state of<br />

pro<strong>to</strong>n could be used as a qubit.<br />

• Distinguish each qubit<br />

Different Larmor frequencies (the<br />

chemical shift) enable us <strong>to</strong> address<br />

each qubit individually.<br />

• <strong>Quantum</strong> logic gates<br />

Radio Frequency (RF) fields and<br />

spin--<br />

--spin couplings between the<br />

nuclei are used <strong>to</strong> implement<br />

quantum logic gates.<br />

82


<strong>Quantum</strong> network and gates<br />

• <strong>Quantum</strong> network<br />

• Entangled gate:<br />

ˆ nπ<br />

J = exp { iγD<br />

⊗ D / 2} , γ = , n = {0,1,... 18}<br />

• The strategy moves U A and U B are<br />

Uˆ<br />

A<br />

⊗Uˆ<br />

B<br />

⎧ Dˆ<br />

⊗ Dˆ ,<br />

⎪<br />

Dˆ<br />

⊗Qˆ,<br />

= ⎨<br />

⎪Q<br />

⊗ Dˆ ,<br />

⎪<br />

⎩ Qˆ<br />

⊗Qˆ,<br />

0 ≤ n ≤ 5<br />

6 ≤ n ≤ 7<br />

,<br />

6 ≤ n ≤ 7<br />

7 ≤ n ≤18<br />

⎛ − ⎞ ⎛ i ⎞<br />

Dˆ<br />

0 1<br />

0<br />

= ⎜ ⎟,<br />

Qˆ<br />

= ⎜ ⎟<br />

⎝−1<br />

0 ⎠ ⎝0<br />

− i⎠<br />

• Each gate can be realized by NMR technique.<br />

36<br />

83


Experiments for quantum game<br />

• Experimentally, we performed nineteen separate sets<br />

of experiments which was distinguished by:<br />

ˆ nπ<br />

J = exp iγD<br />

⊗ D / 2 , γ = , n = {0,1,... 18<br />

{ } }<br />

• In each set, the full process of the game was<br />

executed.<br />

1. Create an effective pure state<br />

2. Prepare the initial entangled state by applying gate J<br />

3. Players Alice and Bob executed their strategic moves U A and U B<br />

4. Apply the unentangled gate J +<br />

5. Measure the final state and calculate the expected payoff.<br />

36<br />

84


NMR Spectrometer<br />

85


Experimental results<br />

• The player Alice's payoffs as a function of the<br />

parameter γ .<br />

• It is easy <strong>to</strong> see that γ = 0 (n=0) corresponds <strong>to</strong><br />

Eisert et al.'s separable game and γ = π 2 (n=18)<br />

corresponds <strong>to</strong> their maximally entangled quantum<br />

game.<br />

86


• Good agreement between theory and experiment.<br />

• Experimental Error:<br />

--an estimated error is less than 0.08, the errors are<br />

primarily due <strong>to</strong> inhomogeneity of magnetic field, imperfect RF<br />

selective pulses, and the variability over time of the mesurement<br />

process.<br />

• Decoherence:<br />

--each experiment <strong>to</strong>ok less than 300 milliseconds, which was<br />

well within the the decoherence time (3 seconds).<br />

• This experiment was referred by :<br />

• Physics News update (APS), Physics web (IOP),<br />

• New Scientist, Science Update (Nature).(<br />

• Physics world<br />

Experimental results<br />

87


2002.4《 Nature 》Science<br />

Update<br />

88


2001.9: APS- Physics News Update<br />

89


2002.1- New Scientists<br />

90


Thanks<br />

91

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!