Introduction to quantum

computing
Miklos Santha
CNRS-LRI Orsay
Overview 1
Basic facts about quantum computing
Grovers search
The hidden subgroup problem
Simons problem
Abelian groups
Shors factorization
Perspectives
History 2
Church Turing thesis (1936), quantitative version
Manin (80) Feynman (82): Simulation of quantum systems
Benio (82) Deutsch (85): QTM
Bennett Brassard (84): Q key distribution
Deutsch (89) Yao (93): Q circuits
Bennett et al. (93): Q teleportation
Bernstein Vazirani (93): Q complexity
Lloyd (93): Q cellular automata
Shor (94): Factorization and discrete logarithm
Grover (96): Q search
Hallgren (02): Pells equation
The qubit 3
Classical bit: b 0, 1
Probabilistic bit
Probability distribution d R
{0,1}
+
such that |d|
1
= 1.
= d = (p, 1 p) with p [0, 1].
Quantum bit
Superposition [) C
{0,1}
such that |[)|
2
= 1.
= [) = [0) + [1) with [[
2
+[[
2
= 1.
[0) =

1
0

, [1) =

0
1

, [) =

.
Qubit evolution 4
Unitary transformation
[) G[), with G C
22
such that G

G = Id.
-
[) G
-
[

)=G[)
Unitary = Reversible:
-
G[) G
-
[)
Measure: Reads and modies.
-
[0)+[1) Measure

:
X
X
Xz
[0)
||
2
[1) ||
2
= Superposition Probability distribution.
Examples 5
Superposition: [) =
1

2
[0) +
1

2
[1)
Measure
- 1

2
[0) +
1

2
[1) Measure

:
X
X
Xz
[0)
1/2
[1) 1/2
Unitary transformations
-
[) G
-
[

)=G[)
NOT, [0) [1): G =

0 1
1 0

.
Hadamard: H =
1

1 1
1 1

.
Quantum coin ip 6
Probabilistic ip
-
0/1 PF

:
X
X
Xz
0
1/2
1 1/2
Remark: PF PF = PF.
Quantum ip
-
[b) H
- 1

2
([0)+(1)
b
[1))

:
X
X
Xz
[0)
1/2
[1) 1/2
Conclusion : PF = Measure H.
Question : H H = ?
Quantum interference 7
: +
1

2
, :
1

2
.
[0)

[0) [1)

[0) [1) [0) [1)
1
2
1
2
1
2

1
2
H H[b) = [b) = H H = Id.
Conclusion : Measures change the computation
The n-qubit 8
Denition: n-qubit tensor product of n qubits.
[) C
{0,1}
n
such that |[)|
2
= 1.
= [) =

x{0,1}
n

x
[x) with

x
[
x
[
2
= 1.
Unitary transformation: [) G[), with G U(2
n
).
-
[) G
-
[

)=G[)
Measure
-

x

x
[x) Measure
-
[x)
|
x
|
2
Partial measure
-
|00 +|01 +|10 +|11 Measure
-
[00) + [10)

[[
2
+[[
2
second bit = 0
Circuits 9
Quantum circuit: (G U(16))
G
H
XOR
R
4
XOR
Theorem [DiV95,BMPRV99]:
Every transformation on n-qubit decomposes into transformations
on 1-qubit and 2-qubit.
=Universal family.
Quantum Fourier transform 10
Circuit
.
.
.
QFT
n .
.
.

H
H
H
.
.
.
Denition
QFT
n
[x) =
1
2
n/2

y
(1)
xy
[y)
where x y =

i
x
i
y
i
mod 2
Quantum computing a function 11
Let f : 0, 1
n
0, 1
m
x f(x)
Reversible
R
f
: 0, 1
n+m
0, 1
n+m
(x, y) (x, y f(x))
Quantum
U
f
U(2
n+m
) : C
2
n+m
C
2
n+m
[x)[y) [x)[y f(x))
QUANTUM SEARCH
Input: f : {0, . . . , N 1} {0, 1}, N = 2
n
f(i) = 1 for some 0 i N 1
f(j) = 0 j = i
Output: i
Complexity : Number of queries to f
Deterministic: N
Probabilistic: N/2
Quantum:
Theorem (Grover): O(

N)
1
GROVERS ALGORITHM
Initialization
|N 1 |i + 1 |i 1 |i |0
1

N
Repeat O(

N) times
Phase shift for |i
|N 1 |i + 1 |i 1 |0
|i
A
Inversion about the average
|N 1 |i + 1 |i 1 |0 |i
A
Amplitude of |i increases by O(
1

N
).
After O(

N) repetitions it becomes O(1).

x{0,1}
n
|x
Phase ip for |i:
Apply U
f
with an ancilla
|x (|0 |1) |x (|0 f(x) |1 f(x))
= (1)
f(x)
|x (|0 |1)
Dispersion:
D = QFT
n
R QFT
n
where
R =
_
_
_
_
_
_
_
_
1 0 0
0 1 0
.
.
.
.
.
.
0 1
_
_
_
_
_
_
_
_
4
ANALYSIS
Recursion for the amplitudes:
r
0
= 1/

N
b
0
= 1/

N
_
_
_
r
k+1
=
N2
N
r
k
+
2(N1)
N
b
k
b
k+1
=
N2
N
b
k

2(N1)
N
r
k
_
_
_
Explicit solution:

sin = 1/

N
r
k
= sin(2k + 1)
b
k
=
1

N1
cos(2k + 1)
_
_
_
Optimum number of iterations:
r
k
= 1 k =
2
4


4

N
5
Simons problem 12
The problem
Input: f : 0, 1
n
0, 1
n
such that
s ,= 0
n
, f(x) = f(y) (x = y or x = y s)
Output: s
Constraint: f is a black box.
Complexity: Number of evaluations of f and computing time.
Deterministically: 2
n1
+ 1 evaluations.
Probabiliste : (2
n/2
) evaluations.
Quantum algorithm [Simon94]: O(n) evaluations and O(n
3
) time.
Idea: Use QFT
n
to nd the period s.
Simons algorithm 13
Circuit
-
-
[0
n
)
[0
n
)
QFT
n
f
QFT
n
Measure
Measure
-
-
?

Analysis
Initialization : [0
n
)[0
n
)
Fourier on 1
st
register:
1
2
n/2

x{0,1}
n
[x) [0
n
)
Evaluation of f :
1
2
n/2

x
[x) [f(x))
Measure of 2
nd
register:
1

2
([x) +[x s)) [f(x))
Fourier on 1
st
register:
1
2
n/2

(1)
xy
+ (1)
(xs)y

[y)
=
1
2
n/2

y
(1)
xy
(1 + (1)
sy
) [y)
Measure of 1
st
register: uniform [y) such that s y = 0
Conclusion
After O(n) iterations, system of rank n1 = 2 solutions : 0
n
, s.
Hidden subgroup problem (HSP) 14
The problem
Input: Finite group G and f : G S which hides H G:
constant and distinct on the left cosets of H.
Output: Generators for H.
G
H
a
1
H
.
.
.
a
t
H
S
Theorem: If G is Abelian then there is a quantum algorithm
which nds H with probability 1 1/[G[,
in polynomial time in log[G[.
Characters of an Abelian group 15
G : Abelian group.
Denition
A character : G C

is a group homorphism.

G = characters of G.
Remark: (x) is a [G[
th
root of the unity.
Theorem: G and

G are isomorphic ;

G =
y
: y G.
Examples : G = Z
d
;
y
(x) =
xy
d
.
G = G
1
G
2
;
y
(x) =
y
1
(x
1
)
y
2
(x
2
).
Orthogonality
H G

H

= y G : h H,
y
(h) = 1.
Lemma. Let H G.
There is a deterministic algorithms which nds H from a set of
generators for H

in time O(log
3
[G[).
Quantum Fourier transform on an Abelian group 16
G : Abelian group.
Bases
Dirac: (
x
)
xG
; [
x
) = [x).
Characters: (
y
)
yG
; [
y
) =

y
(x)[x).
Denition: QFT
G
: [y)
1

G
[
y
).
Principal property. H G, x G.
1

|H|

hH
[x + h) = [x + H)
QFT
G
[H

(x)) =
1

|H

yH

y
(x)[y).
Theorem: Ecient implementation of QFT
G
.
Fourier sampling 17
Circuit : Fourier sampling
f
(G)
-
-
[0)
G
[0)
S
QFT
G
f
QFT
G
Measure
-
?
Analysis
Initialization : [0)
G
[0)
S
Fourier on 1
st
register :
1

|G|

xG
[x) [0)
Query of f :
1

|G|

xG
[x) [f(x)) = [f)
Fourier on 1
st
register :
1
|G|

x,yG

y
(x) [y)[f(x))
=
1

|G|

xG
[0

(x))[f(x))
Measure of 1
st
register : . . .
Fourier sampling when f hides H 18
Circuit : Fourier sampling
f
(G)
-
-
[0)
G
[0)
S
QFT
G
f
QFT
G
Measure
-
H

Analysis
Initialization : [0)
G
[0)
S
Fourier on 1
st
register :
1

|G|

xG
[x) [0)
Query of f : [f) =
1

|G/H|

xG/H
[x + H) [f(x))
Fourier on 1
st
register :
1

|G/H|

xG/H
[H

(x))[f(x))
Measure of 1
st
register : [y) uniform in H

.
Simons algorithm revisited 19
G = 0, 1
n
H = 0
n
, s for some 0
n
,= s 0, 1
n
f(x) = f(y) i x = y or x y = s
The characters

y
: 0, 1
n
C for y 0, 1
n
x (1)
xy
where x y =
n

i=1
x
i
y
i
mod 2
Conclusion
H

= y : s y = 0
n1 independent elements of H

determine H.
Towards factorization 20
Factorization
Input: a composite number n N.
Output: a non trivial divisor of n.
Order nding
Input: n N and a Z

n
.
Output: the period r of x a
x
mod n.
Lemma: Factorization
R
Order nding.
Let a Z
n
random.
Verify that gcd(a, n) = 1.
Compute the period r of x a
x
mod n.
Restart if r is odd or a
r/2
= 1 mod n.
(otherwise (a
r/2
1)(a
r/2
+ 1) = 0 mod n.)
Output gcd(a
r/2
1, n).
Order nding (with help) 21
Input: q, n and a Z

n
such that r = order(a) mod n divides q
Output : r
G = Z
q
, H = 0, r, 2r, . . ., H is a subgroup.
The hiding function:
f : Z
q
Z
n
x a
x
mod n
Characters:

k
: Z
q
C for k Z
q
x e
2i
kx
q

k
(r) = 1 i q/r divides k, H

= k : q/r divides k
Output of Fourier sampling:
Random element of H

: tq/r where t
R
0, 1, . . . r1
If gcd(t, r) = 1 then gcd(tq/r, q) = q/r
Theorem (Shor): The algorithm can be extended to G = Z
QUANTUM COMPLEXITY CLASSES
QP : polynomial time
no error
ZQP : expected polynomial time
no error
BQP : polynomial time
error
1
3
QP
P PSPACE PP ZPP
ZQP BQP
BPP
NP
L = {(x, y) : x has a prime divisor < y} ZQP
Conjecture: L BPP
6
CLASSICAL COMPLEXITY CLASSES
olynomial Time
NPI
NPHard
ntermediate
P
P: Ecient classical algorithms
NPI: Maybe ecient quantum algorithms
NP H: Probably no ecient quantum al-
gorithms
7
NP- INTERMEDIATE PROBLEMS
1. GRAPH ISOMORPHISM
Input: G
1
= (V, E
1
), G
2
= (V, E
2
)
Question : Is there f : V V such that
{u, v} E
1
{f(u), f(v)} E
2
?
2. HIDDEN SUBGROUP
Input: G nite group
: G X
is constant and distinct on the
cosets of a subgroup H of G.
Output : Generating set for H
3. SHORTEST VECTOR
Input: A basis v
1
, . . . , v
n
for a lattice
L Z
n
and w > 0
Question : Is there 0
n
= x L such that
x
2
w?
8
NP- INTERMEDIATE PROBLEMES
4. PIGEONHOLE SUBSET SUM
Input: s
1
, . . . , s
n
N such that

n
i=1
s
i
< 2
n
Output: I
1
= I
2
{1, . . . , n} such that

iI
1
s
i
=

iI
2
s
i
5. IDENTICAL PRODUCT
Input: s
1
, . . . , s
n
N, s
i
< p < 2
n
Output: I
1
= I
2
{1, . . . , n} such that

iI
1
s
i
=

iI
2
s
i
mod p
6. HAPPYNET
Input: G
1
= (V, E
1
) and w : E Z
Output : S : V {1, 1} such that i V
S(i)

{i,j}E
S(j)w(i, j) 0
9