計算モデル論 ー対話証明計算モデルと 量子非局所 ... - imai...

83
計算モデル論 ー対話証明計算モデルと 量子非局所性計算モデルー 今井 2014-06-27, 07-04, 07-09

Upload: others

Post on 14-Feb-2021

5 views

Category:

Documents


0 download

TRANSCRIPT

  • 計算モデル論

    ー対話証明計算モデルと量子非局所性計算モデルー

    今井 浩

    2014-06-27, 07-04, 07-09

  • 背景・動機1

    • 計算量の重要未解決問題– P vs NP

    • 計算モデル(1930’s)– Turing Machine– λ-Calculus– Recursive function

    • Interactive Proof (1985-)

  • • Birch and Swinnerton-Dyer Conjecture • Hodge Conjecture • Navier-Stokes Equations • P vs NP • Poincaré Conjecture • Riemann Hypothesis • Yang-Mills Theory

    http://www.claymath.org/millennium/Birch_and_Swinnerton-Dyer_Conjecture/http://www.claymath.org/millennium/Hodge_Conjecture/http://www.claymath.org/millennium/Navier-Stokes_Equations/http://www.claymath.org/millennium/P_vs_NP/http://www.claymath.org/millennium/Poincare_Conjecture/http://www.claymath.org/millennium/Riemann_Hypothesis/http://www.claymath.org/millennium/Yang-Mills_Theory/

  • 背景・動機2

    • 量子計算・量子情報– Shorの素因数分解量子多項式時間アルゴリズム– Groverの探索アルゴリズム–量子ウォーク

    • 量子暗号– BB84

    • 量子entanglement, 量子非局所性–量子情報処理の力の源⇒対話証明との関係

  • 1927 Solvay Conference on Quantum Mechanicshttp://commons.wikimedia.org/wiki/Image:Solvay_conference_1927.jpgより

    A. Piccard, E. Henriot, P. Ehrenfest, Ed. Herzen, Th. De Donder, E. Schrödinger, E. Verschaffelt, W. Pauli, W. Heisenberg, R.H. Fowler, L. Brillouin,

    P. Debye, M. Knudsen, W.L. Bragg, H.A. Kramers, P.A.M. Dirac, A.H. Compton, L. de Broglie, M. Born, N. Bohr,

    I. Langmuir, M. Planck, M. Curie, H.A. Lorentz, A. Einstein, P. Langevin, Ch. E. Guye, C.T.R. Wilson, O.W. Richardson

    http://commons.wikimedia.org/wiki/Image:Solvay_conference_1927.jpg

  • EPR and Bohr

  • `teleportation’ between distant 2 parties

    Alice Bob

    entangled quantum state

    Alice measures Bob’s statebecomesinstantaneously

    Alice measures Bob’s statebecomesinstantaneously

  • EPR paradox and Bell inequalities• Einstein, Podolsky, Rosen (1935)

    – quantum entanglement vs. relativity theory

    • Bell inequality (1964)– Entanglement/Nonlocalit⇒violation

    • CHSH inequality (Clauser, Horne, Shimony, Holt 1969)– applicable to a bipartite system

    • Aspect et al. (1982)– Experimental verification of violation of

    CHSH inequality

    • Tsirelson (1980): max. violation value⇒ Experimental realization (Sakai et al. 2006)

    entanglement

    measure(local)

    statechange

    instantlyfaster than light

  • Bell-CHSH correlation experiment

    Alice Bob

    entangled quantum state1A

    2A

    1B

    2B

    0)()()()()()( 2212211111 £-+++-- BAPBAPBAPBAPBPAPClassical correlation:

    21)()()()()()( 2212211111 £-+++-- BAPBAPBAPBAPBPAP

    Quantum correlation:

    two measurementsapply one

    two measurementsapply one

  • 対話証明システム

  • Interactive Proof System[Babai 1985; Goldwasser, Micali, and Rackoff 1985]

    • Two players: prover, verifier– Prover tries to convince verifier of her assertion with

    unbounded computational power– Verifier must check validity of prover’s assertion

    probabilistically and efficiently:• probabilistically ⇒ with bounded error• efficiently ⇒ in time polynomial to input length

    Peggy (Prover) Victor (Verifier)Interactive

    Communication

    IP=PSPACE

  • Nondeterministic Polynomial (NP)

    • Two players: prover, verifier– Prover tries to convince verifier of her assertion by

    just given one certificate– Verifier must check validity of prover’s assertion

    efficiently:• efficiently ⇒ in time polynomial to input length

    e?satisfiabl)()()( はzyxzyxzyx ⁄⁄Ÿ⁄⁄Ÿ⁄⁄Peggy (Prover) Victor (Verifier)

    NP

    Oracle

    にして!1,0,1 === zyx

  • Example: Graph Non-Isomorphism

    ◎ Protocol of verifier V:1. Choose an index i Œ {1,2} of graphs

    and a permutation p ΠSn at random.Send a graph p (Gi) to prover P

    to ask which of the two is isomorphic to p (Gi).2. Receive an index j from P.

    Accept iff i = j.

    Graph Non-Isomorphism Problem (GNI)

    INPUT: Two graphs G1, G2 of n vertices

    QUESTION: For all permutation p Œ Sn on vertices,p (G1) π G2?

  • 1

    2 3

    4

    4

    3 1

    2

    同型

    非同型

    1 2

    34

    1

    23

    4

  • 凸多面体入門

  • 7v6v

    凸多面体のV-表現とH-表現

    1v

    4v 5v

    3v2v

    }0,1|{5

    1

    5

    1≥== ÂÂ

    ==ii

    iii

    ivP lll

    I }|{ ii bxaxP £◊=

  • 3点完全グラフK3のcut polytope

    x

    y

    z

    (1,1,0)

    (0,1,1)

    (0,0,0)

    (1,0,1)

    1

    23

    K3

    0£-- zyx

  • 上限定理・線形計画ー Polyhederal Combinatorics ー

    • d 次元 n 点の凸包のファセット数:– (t,t2,t3,…,td) (t=1,2,…,n)の凸包を考えてみて!

    • 線形計画問題

    • 「凸多面体の簡潔な記述≒多項式時間解法」

    • 凸包アルゴリズム:cdd, lrs, etc.–計算幾何(Computational Geometry)

    Î ˚)( 2/dnO

    }0,|min{ ≥=◊ xbAxxc

  • Correlation between 2 Events A,B

    1)()()(

    )()()()(

    1)(),(),(0, Events 2

    £-+

    ££

    ££

    ABPBPAP

    BPABPAPABP

    ABPBPAPBA

    )(Cor 2K□

    Correlation Polytopealso known as Boolean Quadratic Polytope)(AP

    )(BP

    )()(

    BAPABP

    «=

    (1,0,0)

    (0,1,0)

    (1,1,1)B

    A

    A, B

    space-))(),(),(( ABPBPAP

  • Correlation of 2 events

  • Cut polytope of K3

    x

    y

    z

    (1,1,0)

    (0,1,1)

    (0,0,0)

    (1,0,1)

    1

    23

    K3

    )(Cut 2K—=□

    zyx +£triangle inequality

  • Correlation polytope ⇔ Cut polytope

    )(Cor 2K□ )(Cut 2K—

    A B

    )(AP

    X

    )(BP

    )(ABP

    covariance mapping

    23 KK —=

    suspension

  • Correlation of a bipartite system of },{ and },{ 2121 BBAA

    ))(),( no()(),(),(),()(),( ),(),(

    ??? amongn Correlatio

    2121

    22122111

    2121

    BBPAAPBAPBAPBAPBAPBPBPAPAP

    1A 1B)( 11BAP

    2A 2B)( 22BAP

    )( 21BAP )( 12BAP

  • Correlation polytope ⇔ Cut polytope

    )(Cor 2,2K□ )(Cut 2,2K—

    1A 1B

    )( 1AP

    X

    )( 1BP

    )( 11BAP

    covariance mapping

    2,22,2,1 KK —=

    2A 2B)( 22BAP

    )( 21BAP )( 2BP

    suspension

  • Correlation of a bipartite system of},,,{ and },,,{ 2121 BA mm BBBAAA LL

    ))(),( no(),,1;,,1(

    )(),(),(??? amongn Correlatio

    2121 jjii

    BA

    jiji

    BBPAAPmjmi

    BAPBPAPLL ==

    1A 1B)( 11BAP

    AmA

    BmB

    )(BA mm

    BAP

  • Correlation polytope ⇔ Cut polytope

    )(Cor , BA mmK□ )(Cut , BA mmK—

    □covariance mapping

    BABA mmmmKK ,,,1 —=

    )( 1AP

    X

    )( 1BP

    1A 1B)( 11BAP

    AmA

    BmB

    )(BA mm

    BAP

    Problem:Enumerate all the facets of

    )(Cut , BA mmK—□

    to know bipartite correlation

  • 量子対話証明へ

    1対1量子暗号を超えること計算量の枠組みと物理実験の対応

  • Interactive Proof System[Babai 1985; Goldwasser, Micali, and Rackoff 1985]

    • Two players: prover, verifier– Prover tries to convince verifier of her assertion with

    unbounded computational power– Verifier must check validity of prover’s assertion

    probabilistically and efficiently:• probabilistically ⇒ with bounded error• efficiently ⇒ in time polynomial to input length

    Peggy (Prover) Victor (Verifier)Interactive

    Communication

    IP=PSPACE

  • Multi-prover Interactive ProofMIP=NEXPTIME

    Alice Bob

    Victor (Verifier)

    X

  • 量子情報とBell不等式

  • A Two-Party One-Round Interactive Proof System[Cleve, Høyer, Toner, Watrous CCC 2004]

    }1||,,1,0{ -=Œ

    TTt

    L質問

    Alice Bob

    Victor (Verifier)

    Provers

    }1||,,1,0{ -=Œ

    SSs

    L質問

    • 事前に回答戦略を協力して練ってよい• 関数

    • 質問が始まったら通信できない

    }1||,,1,0{ -=Œ

    AAa

    L答

    }1||,,1,0{ -=Œ

    BBb

    L答

    ÔÔÓ

    ÔÔÌ

    Ï

  • 最も単純な場合

    Alice Bob

    Victor (Verifier)

    Provers

    }1,0{=ŒAa答 }1,0{=ŒBb答

    answered] is ),(Pr[)0,0;,( babaC =

  • 相関表t=0

    b=0 b=1s=0

    a=0

    c(0,0) c(0,1)

    a=1

    c(1,0) c(1,1)

    ˙˚

    ˘ÍÎ

    È=

    ==

    0001

    0,0

    c

    ba

    ˙˚

    ˘ÍÎ

    È=

    ==

    0010

    1,0

    c

    ba

    ˙˚

    ˘ÍÎ

    È=

    ==

    01000,1

    c

    ba

    ˙˚

    ˘ÍÎ

    È=

    ==

    10001,1

    c

    ba

    ˙˚

    ˘ÍÎ

    È=

    -===

    -===

    qpqpqppqc

    qqqb

    pppa

    1,]0Pr[

    1,]0Pr[

    0),(

    0),(1

    0,

    =Â=

    jic

    jicji

  • Alice Bob

    Victor (Verifier)

    Provers

    }1,0{Œt質問}1,0{Œs質問

    •事前に回答戦略を協力して練ってよい•質問が始まったら通信できない

    }1,0{Œa答 }1,0{Œb答

    |S|=|T|=2, |A|=|B|=2

  • 相関表(|S|=|T|=2, |A|=|B|=2)

    t=0 t=1b=0 b=1 b=0 b=1

    s=0 a=0 c(0,0;0,0) c(0,1;0,0) c(0,0;0,1) c(0,1;0,1)

    a=1 c(1,0;0,0) c(1,1;0,0) c(1,0;0,1) c(1,1;0,1)

    s=1 a=0 c(0,0;1,0) c(0,1;1,0) c(0,0;1,1) c(0,1;1,1)

    a=1 c(1,0;1,0) c(1,1;1,0) c(1,0;1,1) c(1,1;1,1)

  • 16 deterministic patterns

    ˙˙˙˙

    ˚

    ˘

    ÍÍÍÍ

    Î

    È

    0000010100000101

    ˙˙˙˙

    ˚

    ˘

    ÍÍÍÍ

    Î

    È

    1010000010100000

    ˙˙˙˙

    ˚

    ˘

    ÍÍÍÍ

    Î

    È

    0101000000000101

    ˙˙˙˙

    ˚

    ˘

    ÍÍÍÍ

    Î

    È

    0101000001010000

    ˙˙˙˙

    ˚

    ˘

    ÍÍÍÍ

    Î

    È

    1001000000001001

  • 独立試行相関表

    t=0 t=1b=0 b=1 b=0 b=1

    s=0 a=0 p0q0 p0(1-q0) p0q1 p0(1-q1)

    a=1 (1-p0)q0 (1-p0)(1-q0)

    (1-p0)q1 (1-p0)(1-q1)

    s=1 a=0 p1q0 p1(1-q0) p1q1 p1(1-q1)

    a=1 (1-p1)q0 (1-p1)(1-q0)

    (1-p1)q1 (1-p1)(1-q1)

  • 一般の古典相関表

    ˙˙˙˙

    ˚

    ˘

    ÍÍÍÍ

    Î

    È

    0000010100000101

    ˙˙˙˙

    ˚

    ˘

    ÍÍÍÍ

    Î

    È

    1010000010100000

    ˙˙˙˙

    ˚

    ˘

    ÍÍÍÍ

    Î

    È

    0101000000000101

    ˙˙˙˙

    ˚

    ˘

    ÍÍÍÍ

    Î

    È

    0101000001010000

    ˙˙˙˙

    ˚

    ˘

    ÍÍÍÍ

    Î

    È

    1001000000001001

    確定的相関表の凸一次結合

  • じゃんけんのpay-off matrix

    注:この表では行プレイヤーから見た値で記述

    ゲームの値=0

    グー チョキパーグー 0 1 -1チョキ -1 0 1パー 1 -1 0

  • ゲームの値

    Alice Bob

    Victor (Verifier)

    Provers

    }1||,,1,0{ -=Œ

    TTt

    L質問

    }1||,,1,0{ -=Œ

    SSs

    L質問

    • 事前に回答戦略を協力⇒C(a,b;s,t)• 質問が始まったら通信できない

    }1||,,1,0{ -=Œ

    AAa

    L答

    }1||,,1,0{ -=Œ

    BBb

    L答

    ),;,(),;,(,;,

    tsbaCtsbatsbaWV Â=

    ゲームの値

    ),;.( tsbaW重み

  • A Two-Party One-Round Interactive Proof System[Cleve, Høyer, Toner, Watrous CCC 2004]

    Alice Bob

    Victor (Verifier)

    Provers

    }1||,,1,0{ -=Œ

    TTt

    L質問

    }1||,,1,0{ -=Œ

    SSs

    L質問

    • 事前に回答戦略を協力して練ってよい• 関数

    • 質問が始まったら通信できない

    }1||,,1,0{ -=Œ

    AAa

    L答

    }1||,,1,0{ -=Œ

    BBb

    L答

    ÔÔÓ

    ÔÔÌ

    Ï

  • CHSH Game

    Alice Bob

    Victor (Verifier)

    }1,0{Œt質問}1,0{Œs質問

    }1,0{Œa答 }1,0{Œb答

    ÔÔÓ

    ÔÔÌ

    Ï Ÿ=≈=otherwise0

    1),;,( tsbatsbaW

    ゲームの値:等確率質問に対して証明者が勝つ確率の最大値

    0)1(,1)0(;1)1(,0)0( ==== bbaa4/3 ゲームの値fi

  • CHSH Game

    Alice Bob

    Victor (Verifier)

    }1,0{Œt質問}1,0{Œs質問

    }1,0{Œa答 }1,0{Œb答

    ÔÔÓ

    ÔÔÌ

    Ï Ÿ=≈=otherwise0

    1),;,( tsbatsbaV

    ゲームの値:等確率質問に対して証明者が勝つ確率の最大値

    0)1(,1)0(;1)1(,0)0( ==== bbaa4/3 ゲームの値fi

    s t V0 0 0 1 00 1 0 0 11 0 0 0 11 1 1 1 1

    tsŸ ba≈

    Alice, Bob: prior shared randomness でも3/4は上限

  • pay-off matrix M

    V (0,0¦0,0)(1,1¦1,1)

    (0,0¦0,1)(1,1¦1,0)

    (0,0¦1,0)(1,1¦0,1)

    (0,0¦1,1)(1,1¦0,0)

    (0,1¦0,0)(1,0¦1,1)

    (0,1¦0,1)(1,0¦1,0)

    (0,1¦1,0)(1,0¦0,1)

    (0,1¦1,1)(1,0¦0,0)

    (0,0) 1 1 0 0 1 1 0 0(0,1) 1 0 1 0 1 0 1 0(1,0) 1 1 0 0 0 0 1 1(1,1) 0 1 0 1 1 0 1 0

    ),( ts

    ))1(),0(|)1(),0(( bbaa

    43T0,4

    1,0,41,0,0,4

    1,41

    41,4

    1,41,4

    1: valueGame =˜̃˜˜

    ¯

    ˆ

    ÁÁÁÁ

    Ë

    Ê

    ˜̃˜˜

    ¯

    ˆ

    ÁÁÁÁ

    Ë

    Ê M

    and Optimal Strategies

  • じゃんけんのpay-off matrix

    注:この表では行プレイヤーから見た値で記述

    (前の表では列よりの値)

    グー チョキパーグー 0 1 -1チョキ -1 0 1パー 1 -1 0

  • CHSHゲーム

    • 重み

    ˙˙˙˙

    ˚

    ˘

    ÍÍÍÍ

    Î

    È

    0110100110100101

  • CHSH Gameの解析• 古典の場合: ゲームの値 3/4• 量子の場合:

    – 事前にAliceとBobがエンタングル状態 を共有– それぞれ自分のところで部分測定できる

    1cos0sin)(1

    1sin0cos)(0qqqf

    qqqf

    +-=

    +=

    )8/()8/(1

    )8/()8/(0

    )4/()4/(1

    )0()0(0

    pfpf

    pfpf

    pfpf

    ff

    --=

    =

    =

    =

    bbbY

    bbbY

    aaaXaaaX

    2/)1100( +

    として で測定、結果を回答

    ⇒ゲームの値 75.085.0)8/(2cos >ªp

    まさしくCHSH不等式として知られるBell不等式の話そのもの他にKochen-Specker定理、擬似テレパシー, 量子グラフ彩色など種々の展開

  • 純粋状態での量子情報基礎入門

    ノルム1の複素ベクトルユニタリ行列

    測定

  • 1ビット

    • 古典ビット– “真” か “偽” の2状態のうちどちらかをとる

    • 確率ビット– 確率pで“真”をとり、確率qで“偽”をとる

    – p+q=1 , (0≦ p, q ≦1)

    • 量子ビット– 確率|α|2で“真”をとり、確率|β|2で“偽”をとる

    – |α|2 +|β|2=1 , (α,β は任意の複素数)

    (確率)振幅と呼ぶ10 ba +

  • n ビット

    • 古典 n ビット– 2n 個の状態のうちどれかひとつをとる

    • 量子 n ビット– 2n次元の複素内積空間の単位ベクトル– 2n 個の基底状態の重ね合わせをとる

    12121100 --+++= nn qqq aaay L

    12

    12

    21

    20 =+++= -naaay L

     振幅:,,, 1210 -naaa L

  • entangled 10

    10

    11011000

    1001

    211100

    21

    separable 102

    10

    212

    1

    01

    0011

    210100

    21

    1000

    11

    0100

    10

    0010

    100

    101

    01

    0001

    010

    011

    0000 :)4(qubit 2

    122101

    010

    2 :)2(qubit 1

    :??)|(|

    :)|(||)|(|

    |,|,|,|||

    C,,||||,||,|C

    ˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁ

    Ë

    Ê

    ˜̃˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁ

    Ë

    Ê

    ˜˜˜˜˜˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜̃˜˜˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜˜˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁ

    Ë

    Ê

    ˜̃˜˜˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜̃

    ˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜̃

    ˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜̃

    ˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜˜˜

    ¯

    ˆ

    ÁÁÁ

    Ë

    Ê

    ˜˜˜

    ¯

    ˆ

    ÁÁÁ

    Ë

    Ê

    ˜̃

    ˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜˜˜

    ¯

    ˆ

    ÁÁÁ

    Ë

    Ê

    ˜˜˜

    ¯

    ˆ

    ÁÁÁ

    Ë

    Ê

    ˜˜˜

    ¯

    ˆ

    ÁÁÁ

    Ë

    Ê

    ˜˜˜

    ¯

    ˆ

    ÁÁÁ

    Ë

    Ê

    ˜˜˜

    ¯

    ˆ

    ÁÁÁ

    Ë

    Ê

    ƒ===ÒÒ+

    ÒÒ+ƒÒ=ƒ==ÒÒ+

    =Ò=Ò==Ò==ÒÒƒ=Ò=

    Œ=+=Ò=Ò=Ò=

    gg

    bb

    gbgbgbgb

    bababa

    f

     

    基底 

    N

    N

  • 12111,,10100,0000

    10

    00

    111|,,

    0

    010

    0100|

    ,

    0

    01

    0|0|0|000|2C :qubit

    C,,12||2||,|101|,

    010| :)2(qubit 1

    -=◊◊◊=◊◊◊=◊◊◊

    =Ò◊◊◊=Ò◊◊◊

    =Ò◊ƒ◊◊ƒÒÒƒ=Ò◊◊◊

    Œ=+=Ò=Ò=Ò=

    ˜˜˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜˜˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜˜˜˜˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁÁÁÁÁ

    Ë

    Ê

    ˜˜˜

    ¯

    ˆ

    ÁÁÁ

    Ë

    Ê

    ˜˜˜

    ¯

    ˆ

    ÁÁÁ

    Ë

    Ê

    ˜˜˜

    ¯

    ˆ

    ÁÁÁ

    Ë

    Ê

    n

    nn

    nnnn

    N

    L

    ML

    M

    M

    4847648476

    44444 844444 7648476

    基底

      bababa

    f

  • 量子並列計算

    Ò-+Ò-+

    +Ò+Ò+Ò+Ò=Ò

    --12|22|

    3|2|1 |0||

    1222

    3210nn

    nn aa

    aaaay

    LL

    L

    状態の重ねあわせが可能

    計算過程はそれぞれの状態の振幅が変わっていく過程

    Â-

    =

    =12

    0

    2 1||n

    iia

    → ユニタリ行列をかける

  • 部分測定

    に収斂を測定、全体はで確率

    に収斂を測定、全体はで確率

    で部分測定最左ビットを

     

    yb

    jb

    yjybjby

    ƒÒ

    ƒÒ

    fi

    -ƒÒ+ƒÒ=Ò

    1|1||

    0|0||

    }1,0{

    qubit) 1:,(1 |0||

    21

    20

    10 n

    最左ビットのみ扱える

  • ˜̃˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁ

    Ë

    Ê

    =010

    ˜̃˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁ

    Ë

    Ê

    =101

    ˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁ

    Ë

    Ê

    =ba

    j

    12||2|| =+ ba

    2||a

    2||b

    に状態はで測定を確率

    に状態はで測定を確率

    で測定を直交基底

    1,2||10,2||0

    1,0

    b

    a

    ba

    jǪ̂

    Ô˝¸

    ÔÓ

    ÔÌÏ

    ˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁ

    Ë

    Ê

    =

    射影測定1

  • 0

    1

    ˜˜˜˜˜

    ¯

    ˆ

    ÁÁÁÁÁ

    Ë

    Ê

    +-+

    =ba

    baj2

    1

    12||2|| =+ ba2/2|| ba -

    2/2|| ba +

    に状態はで測定を確率

    に状態はで測定を確率

    で測定を直交基底

    1,2/2||1

    0,2/2||0

    1,0

    ¢-¢

    ¢+¢

    ¢¢Ǫ̂

    Ô˝¸

    ÔÓ

    ÔÌÏ

    ba

    ba

    j

    射影測定20¢1¢

  • ( ) ( )

    ÓÌÏ

    ¢¢==

    ¢¢==

    +=¢+=

    BBBB

    AAAA

    BABA

    BBAA

    00 ,00,00 ,00

    .102

    10,11002

    1

    21

    21

    r

    あまり意味のない測定

    0 1 0 10 ½ ¼ ¼1 ½ ¼ ¼0 ¼ ¼ ½1 ¼ ¼ ½

    1A

    2A

    2B1B

  • ( )

    ( )

    ( ) ( )

    ÔÔÓ

    ÔÔÌ

    Ï

    ==

    ¢¢==

    ÔÔÓ

    ÔÔÌ

    Ï

    ¢++

    =¢++

    =

    +=¢

    +=

    BBBB

    AAAA

    BB

    BABA

    BB

    AA

    ffyy

    yy

    r

    21

    21

    ,

    ,00 ,00

    0122

    1 ,0022

    1

    ,102

    10

    ,11002

    1CHSHに対して有効な測定

    1A

    2A

    2B1B

    0 1 0 10 C D D C1 D C C D0 C D C D1 D C D C

    )8

    (sin21

    2481

    )8

    (cos21

    248223

    2

    2

    p

    p

    =+

    =

    =++

    =

    D

    C

  • CHSH Gameの解析• 古典の場合: ゲームの値 3/4• 量子の場合:

    – 事前にAliceとBobがエンタングル状態 を共有– それぞれ自分のところで部分測定できる

    1cos0sin)(1

    1sin0cos)(0qqqf

    qqqf

    +-=

    +=

    )8/()8/(1

    )8/()8/(0

    )4/()4/(1

    )0()0(0

    pfpf

    pfpf

    pfpf

    ff

    --=

    =

    =

    =

    bbbY

    bbbY

    aaaXaaaX

    2/)1100( +

    として で測定、結果を回答

    ⇒ゲームの値 75.085.0)8/(2cos >ªp

    まさしくCHSH不等式として知られるBell不等式の話そのもの他にKochen-Specker定理、擬似テレパシー, 量子グラフ彩色など種々の展開

  • CHSH is not the only Bell inequality〈A1B1〉+〈A1B2〉+〈A2B1〉-〈A2B2〉 ≤ 2

    -〈A1〉-〈A2〉+〈B1〉+〈B2〉+〈A1B1〉+〈A1B2〉+〈A1B3〉+〈A2B1〉+〈A2B2〉-〈A2B3〉+〈A3B1〉-〈A3B2〉 ≤ 4

    CHSH inequality

    Correlation inequalities with A1,…,A4, B1,…,B4 [Gisin, priv. comm.]

    Tsirelson’s theorem [1980] applicable to inequalities on 〈AiBj〉

    13.514

    Maximum in quantum case:

    (Correlation inequalities; multi-party case consideredby [Werner, Wolf 2001] [Żukowski, Brukner 2002])

    ???5 (2 qubit system) ≤

    I3322 inequality[Pitowsky, Svozil 2001]

    [Collins, Gisin 2004]

    8.165

    2〈A1B1〉+〈A1B2〉+〈A1B3〉+2〈A1B4〉+〈A2B1〉+〈A2B2〉+2〈A2B3〉-2〈A2B4〉+〈A3B1〉+2〈A3B2〉-2〈A3B3〉-〈A3B4〉+2〈A4B1〉-2〈A4B2〉-〈A4B3〉-〈A4B4〉 ≤ 10

    Computed by using SDPA http://grid.r.dendai.ac.jp/sdpa/

    2〈A1B1〉+〈A1B2〉 +〈A1B4〉+〈A2B1〉-〈A2B2〉+〈A2B3〉-〈A2B4〉

    +〈A3B2〉 -〈A3B4〉+〈A4B1〉-〈A4B2〉-〈A4B3〉-〈A4B4〉 ≤ 6

    Most Bell inequalitiesuse some of 〈Ai〉or 〈Bj〉

    Not a correlation inequality

  • An implication of our results

    We give an upper bound for an arbitrary quantum Bell inequality

    Key tools:1. Convex geometry (same as [Tsirelson 1993])

    by solving a semidefinite program

    〈A1B1〉+〈A1B2〉+〈A2B1〉-〈A2B2〉 ≤ 2CHSH inequalityMaximum in quantum case:

    -〈A1〉-〈A2〉+〈B1〉+〈B2〉+〈A1B1〉+〈A1B2〉+〈A1B3〉+〈A2B1〉+〈A2B2〉-〈A2B3〉+〈A3B1〉-〈A3B2〉 ≤ 4

    2. Combinatorial optimization(such as cut polytopes [Deza, Laurent 1997])

    ???5 (2 qubit system) ≤

    I3322 inequality[Pitowsky, Svozil 2001]

    [Collins, Gisin 2004]

  • 量子計算量クラス確率計算量クラス古典計算量クラス

  • Multi-prover Interactive ProofMIP=NEXPTIME

    Alice Bob

    Victor (Verifier)

    X

  • 量子計算量クラス

    P

    NP

    PSPACE

    EXPNEXP

    PZPP

    RP co-RPNP(=EMA) co-NPBPP

    BQP MA=AM1

    QMAAQMA

    PrQP = PP

    BQPSPACE=PrQPSPACE=(N)PSPACE=IP

    EQMARQMA

    NQP=co-C=P

    AM=AM≧2=AM2=IP2

    IP=IPpoly=AMpolyQIP

    EXPNEXP=MIP=QMIP

  • Computation Theoryundecidable

    intractable=exponentialtime

    tractable=polynomialtime

    Halting problem of Turing Machine

    nn log n

    n1.193

    n3.5L log nlog log n

    mediansorting, FFT

    tionmultiplicamatrix nn¥

    linear programming

    Presburger arithmeticcn22

    cn222

    EXPPSPACENP-complete graph isomorphism

    integer factoring

    traveling salesman

    ))3/2)log(log3/1)((logexp( nnO

    decidable

    n: input size

    P

  • 古典計算量クラス

    P

    NP

    PSPACE

    EXPNEXP

    P

    PSPACE

    NEXPEXP

    NP

    Polynomial Time

    NondeterministicPolynomial Time

    Polynomial Space

    Exponential Time

    Tractable

    Intractable

  • 確率計算量クラス

    P

    NP

    PSPACE

    EXPNEXP

    PZPP

    RP co-RPNP co-NPBPP

    PP

    PSPACE

    NEXPEXP

  • • A language L is in PP

    • BPP: Bounded PP• RP: Randomized P• co-RP: complement of RP• ZPP: Zero-error PP

    PP: Probabilistic Polynomial

    ,every for TM, ticprobabilis time-poly: xM$€,2/1] accepts Pr[, if (i) >Œ xMLx.2/1] accepts Pr[, if (ii) £œ xMLx

    1/3 (ii) 2/3, i)( £≥0 (ii) 1/2, i)( =≥

    1/2 (ii) 1, i)( £=0 (ii) 1, i)( ==

  • NP in terms of Probabilistic TMs• A language L is in NP

    ,every for TM, ticprobabilis time-poly: xM$€,0] accepts Pr[, if (i) >Œ xMLx.0] accepts Pr[, if (ii) =œ xMLx

  • 対話計算量クラス

    P

    NP

    PSPACE

    EXPNEXP

    PZPP

    RP co-RPNP co-NPBPP

    MA=AM1

    PP

    PSPACE=IP

    AM=AMc≧2=AM2=IP2

    IP=IPpoly=AMpoly

    EXPNEXP=MIP

  • 量子計算量クラス

    P

    NP

    PSPACE

    EXPNEXP

    PZPP

    RP co-RPNP(=EMA) co-NPBPP

    BQP MA=AM1

    QMAAQMA

    PrQP = PP

    BQPSPACE=PrQPSPACE=(N)PSPACE=IP

    EQMARQMA

    NQP=co-C=P

    AM=AM≧2=AM2=IP2

    IP=IPpoly=AMpolyQIP

    EXPNEXP=MIP=QMIP

  • FPTAS, FPRAS

  • Quantum Adiabatic Computation,Quantum Annealing

    and Ising Partition Function

  • Quantum adiabatic computation/quantum annealing

    GroundState

    GroundState

    GroundState

    Quantum AnnealingHamiltonian

    in the Ising model

    Almost Time-Reversible

    Hamiltonian witheasily prepared state

  • Ising Model+1

    +1

    -1+1

    -1

    -1

  • Ising model [Ising 25]• グラフ𝐺𝐺 = 𝑉𝑉,𝐸𝐸 , 𝑣𝑣 ∈ 𝑉𝑉 :点, 𝑒𝑒 = (𝑢𝑢, 𝑣𝑣)(∈ 𝐸𝐸): 枝• スピン 𝜎𝜎 ∈ {−1,1}𝑉𝑉, 相互作用力 𝐽𝐽𝑢𝑢𝑢𝑢, 外部磁場𝑀𝑀𝑢𝑢

    Hamiltonian 𝐻𝐻(𝜎𝜎) = − �(𝑢𝑢,𝑢𝑢)∈𝐸𝐸

    𝐽𝐽𝑢𝑢𝑢𝑢𝜎𝜎𝑢𝑢𝜎𝜎𝑢𝑢 −�𝑢𝑢∈𝑉𝑉

    𝑀𝑀𝑢𝑢𝜎𝜎𝑢𝑢

    Partition function 𝑍𝑍 𝐺𝐺,𝛽𝛽 = �𝜎𝜎∈{−1,1}𝑉𝑉

    𝑒𝑒−𝛽𝛽𝐻𝐻 𝜎𝜎

    ÿ𝐽𝐽𝑢𝑢𝑢𝑢 > 0: ferromagnetic, 強磁性ÿ𝐽𝐽𝑢𝑢𝑢𝑢 < 0: antiferromagnetic, 反強磁性

  • Simulated Annealing

    • Proposed byS. Kirkpatrick, C. D. Gelatt, Jr, and M. P. Vecchi: Optimization by Simulated Annealing. Science, Vol.220, No.4598 (1983), pp.671–680.

    • Considered applying the Metropolis’ algorithm to the Ising model in the beginning,

  • Quantitative Church-Turing Thesis

    • Thesis 1.1 (quantitative Church’s thesis). Any physical computing device can be simulated by a Turing machine in a number of steps polynomial in the resources used by the computing device.[Short, SICOMP 1997]

  • Scott Aaronson’s Research Statement

    … Either

    1. the Extended Church-Turing Thesis is false,

    2. quantum mechanics as conventionally understood is false, or

    3. the factoring problem is solvable in polynomial time on a classical computer.

    http://www.scottaaronson.com/research.pdf

    80

  • arXiv:1212.1739

    arXiv:1305.4904

    arXiv:1401.7087

    arXiv:1404.6499

    arXiv:1305.5837

    arXiv:1304.4595Nature Phys. 10, 218 (2014)

    arXiv:1403.4228

    Series of papers

    arXiv:1401.2910Science 1252319, Published online 19 June 2014.

  • レポート問題

    1. Bell不等式について、以下の問に答えよ。1. 凸包ソフトウェア(cdd, lrsなど)を用いて、一般の

    Bell不等式を求め、考察を加えよ。2. Correlation polytopeとCut polytopeそれぞれが、Bell不等式の形にどう影響するか。具体的にcovariance mapを与えて議論せよ。

    2. 次の論文を読んで議論せよ。Scott Aaronson: NP-complete problems and physical reality. ACM SIGACT News, Vol.36, No.1 (2005), pp.30-52.

  • 提出

    • レポートは1問のみ回答でよい• 理7号館1階情報科学科レポートボックス• 締切:8月末(できる限り7月中旬に)

    計算モデル論��ー対話証明計算モデルと�量子非局所性計算モデルー背景・動機1スライド番号 3背景・動機21927 Solvay Conference on Quantum Mechanics �http://commons.wikimedia.org/wiki/Image:Solvay_conference_1927.jpgよりEPR and Bohr`teleportation’ between distant 2 partiesEPR paradox and Bell inequalitiesBell-CHSH correlation experiment対話証明システムInteractive Proof System� [Babai 1985; Goldwasser, Micali, and Rackoff 1985]Nondeterministic Polynomial (NP)Example: Graph Non-Isomorphismスライド番号 14凸多面体入門凸多面体のV-表現とH-表現3点完全グラフK3のcut polytope上限定理・線形計画�ー Polyhederal Combinatorics ーCorrelation between 2 Events A,BCorrelation of 2 eventsCut polytope of K3Correlation polytope ⇔ Cut polytopeCorrelation of a bipartite system ofCorrelation polytope ⇔ Cut polytopeCorrelation of a bipartite system ofCorrelation polytope ⇔ Cut polytope量子対話証明へInteractive Proof System� [Babai 1985; Goldwasser, Micali, and Rackoff 1985]Multi-prover Interactive Proof量子情報とBell不等式A Two-Party One-Round Interactive Proof System�[Cleve, Høyer, Toner, Watrous CCC 2004]最も単純な場合相関表|S|=|T|=2, |A|=|B|=2相関表(|S|=|T|=2, |A|=|B|=2)16 deterministic patterns独立試行相関表一般の古典相関表じゃんけんのpay-off matrixゲームの値A Two-Party One-Round Interactive Proof System�[Cleve, Høyer, Toner, Watrous CCC 2004]CHSH GameCHSH Gamepay-off matrix Mじゃんけんのpay-off matrixCHSHゲームCHSH Gameの解析純粋状態での量子情報基礎入門1ビット n ビットスライド番号 51スライド番号 52量子並列計算部分測定射影測定1射影測定2あまり意味のない測定CHSHに対して有効な測定CHSH Gameの解析CHSH is not the only Bell inequalityAn implication of our results�量子計算量クラス�確率計算量クラス�古典計算量クラスMulti-prover Interactive Proof量子計算量クラスComputation Theory古典計算量クラス確率計算量クラスPP: Probabilistic PolynomialNP in terms of Probabilistic TMs対話計算量クラス量子計算量クラスFPTAS, FPRASQuantum Adiabatic Computation,�Quantum Annealing�and �Ising Partition FunctionQuantum adiabatic computation/quantum annealingIsing ModelIsing model [Ising 25]Simulated Annealingスライド番号 78Quantitative Church-Turing ThesisScott Aaronson’s Research StatementSeries of papersレポート問題提出