Download The Hidden Abelian Subgroup Problem

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
The Hidden Subgroup Problem
πΈπ‘™π‘’π‘“π‘‘β„Žπ‘’π‘Ÿπ‘–π‘œπ‘  π‘€π‘œπ‘ π‘β„Žπ‘Žπ‘›π‘‘π‘Ÿπ‘’π‘œπ‘’
The Hidden Subgroup Problem
Problem of great importance in Quantum Computation
β€’
β€’
Most Q.A. that run exponentially faster than their classical
counterparts fall into the framework of HSP
Simon’s Algorithm , Shor’s Algorithm for factoring , Shor’s discrete
logarithm algorithm equivalent to HSP
Quantum Fourier Transform
Discrete Fourier Transform , maps the sequence of complex numbers π‘₯0 , π‘₯1 , … , π‘₯π‘βˆ’1
onto an N periodic sequence of complex numbers 𝑋0 , 𝑋1 , … , π‘‹π‘βˆ’1
π‘βˆ’1
π‘‹π‘˜ =
π‘₯𝑛 β‹…
𝑛=0
2πœ‹πš€
βˆ’
𝑒 𝑁 β‹…π‘˜π‘›
π‘βˆ’1
π‘₯𝑛 β‹… πœ”βˆ’π‘˜π‘›
=
𝑛=0
Quantum Fourier Transform
Discrete Fourier Transform , maps the sequence of complex numbers π‘₯0 , π‘₯1 , … , π‘₯π‘βˆ’1
onto an N periodic sequence of complex numbers 𝑋0 , 𝑋1 , … , π‘‹π‘βˆ’1
π‘βˆ’1
π‘‹π‘˜ =
π‘₯𝑛 β‹…
2πœ‹πš€
βˆ’
𝑒 𝑁 β‹…π‘˜π‘›
π‘βˆ’1
π‘₯𝑛 β‹… πœ”βˆ’π‘˜π‘›
=
𝑛=0
𝑛=0
Quantum Fourier Transform , acts on a quantum state
it in the quantum state
|π‘˜ =
π‘βˆ’1
π‘˜=0 π‘₯π‘˜ β‹…
π‘βˆ’1
1
βˆšπ‘ 𝑛=0
π‘βˆ’1
𝑛=0 π‘₯𝑛
β‹… |𝑛 and transforms
|π‘˜
|𝑛 β‹…
2πœ‹πš€
βˆ’
𝑒 𝑁 β‹…π‘˜π‘›
π‘βˆ’1
|𝑛 β‹… πœ”βˆ’π‘˜π‘›
=
𝑛=0
Quantum Fourier Transform
QFT as a unitary matrix:
𝐹𝑁 =
1
π‘βˆ’1 π‘βˆ’1
βˆšπ‘ 𝑛=0 β„“=0
2πœ‹πš€
βˆ’
𝑒 𝑁 β‹…π‘˜β„“ |β„“
Can implemented in a quantum circuit as a set
of Hadamard and phase shift gates.
𝑛(𝑛+1)
2
gates
π‘˜|
Quantum Fourier Transform
QFT as a unitary matrix:
𝐹𝑁 =
1
π‘βˆ’1 π‘βˆ’1
βˆšπ‘ 𝑛=0 β„“=0
2πœ‹πš€
βˆ’
𝑒 𝑁 β‹…π‘˜β„“ |β„“
Can implemented in a quantum circuit as a set
of Hadamard and phase shift gates.
𝑛(𝑛+1)
2
gates
Example 3 qubit QFT:
π‘˜|
Shor’s Algorithm
Purpose: Factor an Integer 𝑁0
Shor’s Algorithm
Purpose: Factor an Integer 𝑁0 (e.g.
𝑁0 = 21)
1. Choose a random integer a (e.g. π‘Ž = 11)
2. Define a function 𝑓 π‘₯ = π‘Ž π‘₯ π‘šπ‘œπ‘‘π‘’π‘™π‘œ 𝑁0 : 1,11,16,8,4,2,1,11,16
Shor’s Algorithm
Purpose: Factor an Integer 𝑁0 (e.g.
𝑁0 = 21)
1. Choose a random integer a (e.g. π‘Ž = 11)
2. Define a function 𝑓 π‘₯ = π‘Ž π‘₯ π‘šπ‘œπ‘‘π‘’π‘™π‘œ 𝑁0 : 1,11,16,8,4,2,1,11,16
Can be implemented by the Quantum Circuit:
Shor’s Algorithm
1. |πœ“0 =|0
βŠ—π‘›
|0
βŠ—π‘›
Shor’s Algorithm
1. |πœ“0 =|0
βŠ—π‘›
|0
βŠ—π‘›
2. |πœ“1 = 𝐻
βŠ—π‘›
|0
βŠ—π‘›
|0
βŠ—π‘›
=
1
βˆšπ‘
π‘₯∈2𝑛
|π‘₯ |0
βŠ—π‘›
Shor’s Algorithm
1. |πœ“0 =|0
2. |πœ“1 = 𝐻
3. |πœ“2 =
βŠ—π‘›
βŠ—π‘›
|0
|0
1
π‘ˆπ‘“
βˆšπ‘
βŠ—π‘›
βŠ—π‘›
|0
π‘₯∈2𝑛 |π‘₯
βŠ—π‘›
|0
=
1
βˆšπ‘
βŠ—π‘›
=
π‘₯∈2𝑛
1
βˆšπ‘
|π‘₯ |0
π‘₯∈2𝑛 |π‘₯
βŠ—π‘›
|𝑓(π‘₯)
Shor’s Algorithm
1. |πœ“0 =|0
2. |πœ“1 = 𝐻
3. |πœ“2 =
4. |πœ“3 =
βŠ—π‘›
βŠ—π‘›
|0
|0
βŠ—π‘›
βŠ—π‘›
|0
1
π‘ˆπ‘“
𝑛 |π‘₯
π‘₯∈2
βˆšπ‘
1
π‘šβˆ’1
|π‘₯0 +
π‘˜=0
βˆšπ‘š
βŠ—π‘›
|0
=
1
βˆšπ‘
βŠ—π‘›
=
π‘₯∈2𝑛
1
βˆšπ‘
|π‘₯ |0
π‘₯∈2𝑛 |π‘₯
βŠ—π‘›
|𝑓(π‘₯)
π‘˜ β‹… π‘Ÿ |𝑓(π‘₯0 )
First register collapses into a superposition of the preimages of 𝑓(π‘₯)
Shor’s Algorithm
4. |πœ“3 =
1
βˆšπ‘š
π‘šβˆ’1
π‘˜=0 |π‘₯0
+π‘˜β‹…π‘Ÿ
Restrict the study in the domain 𝑍𝑁 ≀ 𝑍 with N a multiple of the period 𝑁 = π‘š β‹… π‘Ÿ
5. |πœ“4 =
1
π‘ˆπ‘„πΉπ‘‡
βˆšπ‘š
π‘šβˆ’1
π‘˜=0 |π‘₯0
+π‘˜β‹…π‘Ÿ
Shor’s Algorithm
4. |πœ“3 =
1
βˆšπ‘š
π‘šβˆ’1
π‘˜=0 |π‘₯0
+π‘˜β‹…π‘Ÿ
Restrict the study in the domain 𝑍𝑁 ≀ 𝑍 with N a multiple of the period 𝑁 = π‘š β‹… π‘Ÿ
5. |πœ“4 =
1
π‘ˆπ‘„πΉπ‘‡
βˆšπ‘š
π‘šβˆ’1
π‘˜=0 |π‘₯0
+π‘˜β‹…π‘Ÿ
𝐹𝑁 =
1
π‘βˆ’1 π‘βˆ’1
βˆšπ‘ 𝑗=0
𝑖=0
2πœ‹πš€
βˆ’ 𝑁 ⋅𝑗𝑖
𝑒
|𝑗 𝑖|
Shor’s Algorithm
4. |πœ“3 =
1
βˆšπ‘š
π‘šβˆ’1
π‘˜=0 |π‘₯0
+π‘˜β‹…π‘Ÿ
Restrict the study in the domain 𝑍𝑁 ≀ 𝑍 with N a multiple of the period 𝑁 = π‘š β‹… π‘Ÿ
5. |πœ“4 =
1
π‘ˆπ‘„πΉπ‘‡
βˆšπ‘š
|πœ“π‘“ =
1
π‘šβˆ’1
π‘˜=0 |π‘₯0
π‘βˆ’1
βˆšπ‘Ÿ 𝑗:π‘šπ‘’π‘™π‘‘π‘–π‘π‘™π‘’ π‘œπ‘“ π‘š
+π‘˜β‹…π‘Ÿ
𝐹𝑁 =
2πœ‹πš€
βˆ’ 𝑁 β‹…π‘₯0 𝑗
𝑒
|𝑗
1
π‘βˆ’1 π‘βˆ’1
βˆšπ‘ 𝑗=0
𝑖=0
2πœ‹πš€
βˆ’ 𝑁 ⋅𝑗𝑖
𝑒
|𝑗 𝑖|
Shor’s Algorithm
|πœ“π‘“ =
1
π‘βˆ’1
βˆšπ‘Ÿ 𝑗:π‘šπ‘’π‘™π‘‘π‘–π‘π‘™π‘’ π‘œπ‘“ π‘š
2πœ‹πš€
βˆ’ 𝑁 β‹…π‘₯0 𝑗
𝑒
|𝑗
Perform measurement: get a j (and thus a multiple of m)
After k trials obtain k number multiples of m.
Shor’s Algorithm
|πœ“π‘“ =
1
π‘βˆ’1
βˆšπ‘Ÿ 𝑗:π‘šπ‘’π‘™π‘‘π‘–π‘π‘™π‘’ π‘œπ‘“ π‘š
2πœ‹πš€
βˆ’ 𝑁 β‹…π‘₯0 𝑗
𝑒
|𝑗
Perform measurement: get a j (and thus a multiple of m)
After k trials obtain k number multiples of m.
𝑁
π‘š = 𝐺𝐢𝐷(𝑗1 , 𝑗2 , … , π‘—π‘˜ ) . It is π‘Ÿ = . Period is found !
π‘š
π‘Ž0 = 1 β†’ π‘Žπ‘Ÿ = 1 β†’ π‘Žπ‘Ÿ/2 + 1 π‘Žπ‘Ÿ/2 βˆ’ 1 = 0
π‘šπ‘œπ‘‘(𝑁0 )
Shor’s Algorithm
|πœ“π‘“ =
1
π‘βˆ’1
βˆšπ‘Ÿ 𝑗:π‘šπ‘’π‘™π‘‘π‘–π‘π‘™π‘’ π‘œπ‘“ π‘š
2πœ‹πš€
βˆ’ 𝑁 β‹…π‘₯0 𝑗
𝑒
|𝑗
Perform measurement: get a j (and thus a multiple of m)
After k trials obtain k number multiples of m.
𝑁
π‘š = 𝐺𝐢𝐷(𝑗1 , 𝑗2 , … , π‘—π‘˜ ) . It is π‘Ÿ = . Period is found !
π‘š
π‘Ž0 = 1 β†’ π‘Žπ‘Ÿ = 1 β†’ π‘Žπ‘Ÿ/2 + 1 π‘Žπ‘Ÿ/2 βˆ’ 1 = 0
π‘šπ‘œπ‘‘(𝑁0 )
One of the factors may has a common factor with 𝑁0
Elements of Group Theory
Group G: set of elements {g} , equipped with an internal composition law
Elements of Group Theory
Group G: set of elements {g} , equipped with an internal composition law
βˆ€π‘Ž, 𝑏 ∈ 𝐺 β†’ π‘Ž ∘ 𝑏 = 𝑐 ∈ 𝐺
π‘Žβˆ˜π‘ βˆ˜π‘ =π‘Žβˆ˜ π‘βˆ˜π‘
Identity element e: 𝑒 ∘ π‘Ž = π‘Ž ∘ 𝑒 = π‘Ž
Inverse element βˆ€π‘Ž ∈ 𝐺 βˆƒ π‘Žβˆ’1 : π‘Žβˆ’1 ∘ π‘Ž = π‘Ž ∘ π‘Žβˆ’1 = 𝑒
Elements of Group Theory
Group G: set of elements {g} , equipped with an internal composition law
βˆ€π‘Ž, 𝑏 ∈ 𝐺 β†’ π‘Ž ∘ 𝑏 = 𝑐 ∈ 𝐺
π‘Žβˆ˜π‘ βˆ˜π‘ =π‘Žβˆ˜ π‘βˆ˜π‘
Identity element e: 𝑒 ∘ π‘Ž = π‘Ž ∘ 𝑒 = π‘Ž
Inverse element βˆ€π‘Ž ∈ 𝐺 βˆƒ π‘Žβˆ’1 : π‘Žβˆ’1 ∘ π‘Ž = π‘Ž ∘ π‘Žβˆ’1 = 𝑒
If π‘Ž ∘ 𝑏 = 𝑏 ∘ π‘Ž : Abelian Group
Subgroup: a non empty set which is a group on its own, under the same composition law
Elements of Group Theory
Group G: set of elements {g} , equipped with an internal composition law
βˆ€π‘Ž, 𝑏 ∈ 𝐺 β†’ π‘Ž ∘ 𝑏 = 𝑐 ∈ 𝐺
π‘Žβˆ˜π‘ βˆ˜π‘ =π‘Žβˆ˜ π‘βˆ˜π‘
Identity element e: 𝑒 ∘ π‘Ž = π‘Ž ∘ 𝑒 = π‘Ž
Inverse element βˆ€π‘Ž ∈ 𝐺 βˆƒ π‘Žβˆ’1 : π‘Žβˆ’1 ∘ π‘Ž = π‘Ž ∘ π‘Žβˆ’1 = 𝑒
If π‘Ž ∘ 𝑏 = 𝑏 ∘ π‘Ž : Abelian Group
Subgroup: a non empty set which is a group on its own, under the same composition law
Cosets: H a subgroup of G. Choose an element g. The (left) coset of H in terms of g is
𝑔𝐻 = π‘”β„Ž | β„Ž ∈ 𝐻
Two cosets of H can either totally match or be totally different
The Hidden Abelian Subgroup Problem
Let G be a group , H a subgroup and X a set.
Let 𝑔1 𝐻 = 𝑔2 𝐻 . A function 𝑓 𝑔 separates the cosets of H iff 𝑓 𝑔1 = 𝑓 𝑔2 .
The function 𝑓 separates the cosets.
The Hidden Abelian Subgroup Problem
Let G be a group , H a subgroup and X a set.
Let 𝑔1 𝐻 = 𝑔2 𝐻 . A function 𝑓 𝑔 separates the cosets of H iff 𝑓 𝑔1 = 𝑓 𝑔2 .
The function 𝑓 separates the cosets.
HSP: determine the subgroup H using information gained by the evaluation of 𝑓.
Assume that elements of G are encoded to basis states of a Quantum Computer.
Assume that exists a β€œblack box” that performs π‘ˆπ‘“
The Hidden Abelian Subgroup Problem
The Simplest Example
Let 𝐺 = 𝑍𝑁 , + e.g. 𝑍3 = 0,1,2
𝑓: 𝐺 β†’ 𝑋 separates cosets
𝐻 = 𝑑 and 𝐻 = 𝑀
The Hidden Abelian Subgroup Problem
The Simplest Example
Let 𝐺 = 𝑍𝑁 , + e.g. 𝑍3 = 0,1,2
𝑓: 𝐺 β†’ 𝑋 separates cosets
𝐻 = 𝑑 and 𝐻 = 𝑀
We don’t know M, d, H but we know G and we have a
β€œmachine” performing the function f
The Hidden Abelian Subgroup Problem
The Simplest Example
Map: 𝐺: |0 , |1 , |2 , … , |𝑁 βˆ’ 1
𝐻: |0 , |𝑑 , |2𝑑 , … , | 𝑀 βˆ’ 1 𝑑
Quantum circuit:
The Hidden Abelian Subgroup Problem
1. |πœ“0 =|0 |0
The Hidden Abelian Subgroup Problem
1. |πœ“0
=|0 |0
2. |πœ“1 = UQFT |0 |0 =
1
βˆšπ‘
π‘βˆ’1
|𝑗 |0
𝑗=0
The Hidden Abelian Subgroup Problem
1. |πœ“0
=|0 |0
2. |πœ“1 = UQFT |0 |0 =
1
βˆšπ‘
π‘βˆ’1
|𝑗 |0
𝑗=0
3. |πœ“2 = π‘ˆπ‘“
1
βˆšπ‘
π‘βˆ’1
|𝑗 |0 =
𝑗=0
1
βˆšπ‘
π‘βˆ’1
|𝑗 |𝑓(𝑗)
𝑗=0
Measure the second register. The function acquires a certain value 𝑓(𝑗0 ). The first register has to
collapse to those j that belong to the coset of H. Entanglement : computational speed up.
The Hidden Abelian Subgroup Problem
|πœ“3 =
1
βˆšπ‘€
|𝑗0 + β„Ž |𝑓(𝑗0 ) =
β„Žβˆˆπ»
|πœ“4 =
1
𝑛
βˆšπ‘‘ 𝑖=0
1
βˆšπ‘€
2πœ‹πš€π‘—0
𝑒 𝑁 ⋅𝑑𝑀
π‘€βˆ’1
|𝑗0 + 𝑠𝑑 |𝑓(𝑗0 )
𝑠=0
|𝑑𝑀
The Hidden Abelian Subgroup Problem
|πœ“3 =
1
βˆšπ‘€
|𝑗0 + β„Ž |𝑓(𝑗0 ) =
β„Žβˆˆπ»
|πœ“4 =
1
𝑛
βˆšπ‘‘ 𝑖=0
1
βˆšπ‘€
2πœ‹πš€π‘—0
𝑒 𝑁 ⋅𝑑𝑀
π‘€βˆ’1
|𝑗0 + 𝑠𝑑 |𝑓(𝑗0 )
𝑠=0
|𝑑𝑀
A measurement will yield a value for M. Repeat and use Euclidean algorithm for the GCD to find
M. Since 𝑁 = 𝑀 β‹… 𝑑 the generating set can be determined and thus H.
The Hidden Abelian Subgroup Problem
|πœ“3 =
1
βˆšπ‘€
|𝑗0 + β„Ž |𝑓(𝑗0 ) =
β„Žβˆˆπ»
|πœ“4 =
1
𝑛
βˆšπ‘‘ 𝑖=0
1
βˆšπ‘€
2πœ‹πš€π‘—0
𝑒 𝑁 ⋅𝑑𝑀
π‘€βˆ’1
|𝑗0 + 𝑠𝑑 |𝑓(𝑗0 )
𝑠=0
|𝑑𝑀
A measurement will yield a value for M. Repeat and use Euclidean algorithm for the GCD to find
M. Since 𝑁 = 𝑀 β‹… 𝑑 the generating set can be determined and thus H.
References
Chris Lomont: http://arxiv.org/pdf/quant-ph/0411037v1.pdf
Frederic Wang http://arxiv.org/ftp/arxiv/papers/1008/1008.0010.pdf
http://en.wikipedia.org/wiki/Quantum_Fourier_transform
Related documents