Survey
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
Quantum information Theory: Separability and distillability J. Ignacio Cirac Institute for Theoretical Physics University of Innsbruck KIAS, November 2001 SFB Coherent Control €U TMR Entangled states Superposition principle in Quantum Mechanics: If the systems can be in j0i or j1i then they can also be in Two or more systems: A B c j 0 i + c j 1 i 0 1 entangled states If the systems can be in j 0 iA j 0 iB or j 1 iA j 1 iB j 0 i j 0 i + c j 1 i j 1 i then they can also be in c 0 A B 1 A B Entangled states possess non-local (quantum) correlations: A B The outcomes of measurements in A and B are correlated. In order to explain these correlations classically (with a realistic theory), we must have non-locality. Fundamental implications: Bell´s theorem. Applications Secret communication. Alice Bob j Á i = j 0 ; 1 i + j 1 ; 0 i 1. Check that particles are indeed entangled. Correlations in all directions. No eavesdropper present 2. Measure in A and B (z direction): Alice Bob 0 1 1 1 0 0 1 1 1 0 Send secret messages Given an entangled pair, secure secret communication is possible Computation. ouput quantum processor input jªouti j ª i = U j ª i o u t i n jªini A quantum computer can perform ceratin tasks more efficiently A quantum computer can do the same as a classical computer ... and more - Factorization (Shor). - Database search (Grover). - Quantum simulations. Precission measurements: We can measure more precisely Efficient communication: 1 Alice 101 1 Alice 101 + 0 0 1 Bob 0 0 1 Bob Entangled state We can use less resources Problem: Decoherence environment A B The systems get entangled with the environment. j © i j E i ! j ª i A B E A B E Reduced density operator: ½ = t r ( j ª i h ª j ) = 6 j © i h © j E A B Solution: Entanglement distillation ½ Idea: ½ ½ ... local operation ½ environment (classical communication) Distillation: j 0 ; 1 i + j 1 ; 0 i local operation ... Fundamental problems in Quantum Infomation: Separability and distillability SEPARABILITY A ½ B DISTILLABILITY ½ ½ ... ½ j © i = j 0 ; 1 i + j 1 ; 0 i ½ Are these systems entangled? Can we distill these systems? Additional motivations: Experiments Separability: Ion traps Cavity QED Optical lattices Magnetic traps Atomic ensembles Distillability: quantum communication. jÁi Long distance Q. communication? NMR Quantum dots Josephson junctions Basic properties: Th. Physics Mathematics Separability This talk Distillability Quantum Information Algorithms, etc: Computer Science Physical implementations: Th. Physics Exp. Physics Q. Optics Condensed Matter NMR Outline Separability. Distillability. Gaussian states. Separability. Distillability. Multipartite case: 1. Separability ½ Are these systems entangled? 1.1 Pure states Product states are those that can be written as: j © i = j a i j b i Otherwise, they are entangled. Examples: Product state: j 0 iA j 0 iB j 0 i j 0 i + c j 1 i j 1 i Entangled state: c 0 A B 1 A B Entangled states cannot be created by local operations. j a i j b i ! j a ( t ) i j b ( t ) i 1.2 Mixed states ½ In order to create an entangled state, one needs interactions. Separable states are those that can be prepared by LOCC out of a product state. Otherwise, they are entangled. X ¸0 = p j a i h a j j b i h b jwhere p k A state is separable iff ½ k k k k k (Werner 89) k Problem: given ½, there are infinitely many decompositions X ½= ¸ kjª k i hª k j ª j ª i = ± spectral decomposition h k j k ; j qk j©k i h©kj © j © i = 6 ± k j k ; j need not be orthogonal h k X = k = ::: X ½= pkjak i hak j - jbk i hbkj k 22 = C C Example: two qubits ( H ) 00 01 10 11 0 1 7122 1B 1322C B C ½ = @ 2 0 2231A 2217 1 2 ½= (j0; 0i h0; 0j + j1; 1i h1; 1j) + j+ ; + i h+ ; + j 5 5 1 + (j0; ¡ i h0; ¡ j + j1; ¡ i h1; ¡ j) 10 where 1 p j § i= ( j 0 i§ j 1 i ) 2 1.3 Separability: positive maps A linear map L :A ( H ) A ( H )is called positive ½ ¸0 !L ( ½ ) ¸0 ½ ½ A B state Extensions state A A ½ B B state ? : need not be positive, in general 0 ! ( L 1 ) ( ½ ) ¸ 0 A postive map is completely positive if: ½ A B̧ A B ½ is separable iff ( for all positive maps L 1 ) ( ½ ) ¸ 0 (Horodecki 96) However, we do not know how to construct all positive maps. Example: Any physical action. ½ ¸0 L ( ½ ) ¸0 A A ½ B B state state Any physical action can be described in terms of a completely positive map. Example: transposition (in a given basis) X ½ = ½ iih jj i ; jj i ; j 1 5 X T ( ½ ) = ½ j j i h i j i ; j µ 2 1¡ i 1+ i 3 ; ¶ 1 5 ! µ 2 1+ i 1¡ i 3 ¶ i ; j Is positive A A B B Extension: partial transposition. X X Is called ½ = ½ j i ; j i h k ; l j ( T 1 ) ( ½ ) = ½ j k ; j i h i ; l j , then i j k l i j k l 0 1 B B @ C C A transposes the blocks partial transposition Example: 0 1 B 1 B0 0 2@ 1 0 0 0 0 0 0 0 0 1 0 1 1 0 C B 0 C 1 B0 0 ! 0 A 2@ 0 1 1 0 0 Partial transposition is positive, but not completely positive. 0 1 0 0 1 0 0C C 0A 1 What is known? PPT In general NPT - Low rank - Necessary or sufficient conditions ? 2x2 and 2x3 Gaussian states (Horodecki and Peres 96) PPT (Horodecki 97) NPT (Giedke, Kraus, Lewenstein, Cirac, 2001) PPT NPT 2. Distillability ½ ½ ... ½ j © i = j 0 ; 1 i + j 1 ; 0 i Can we distill MES using LOCC? PPT states cannot be distilled. Thus, there are bound entangled states. (Horodecki 97) There seems to be NPT states that cannot be distilled. (DiVincezo et al, Dur et al, 2000) 2.1 NPT states (IBM, Innsbruck 99) We just have to concentrate on states with non-positive partial transposition. T A ¸ 0then there exists A and B, such that Idea: If ½ · Z Ţ A y y ½ = ( A B ) ½ ( A B ) with ~ d ¹ ( U U ) ½ ~ ( U U )¸ 0 U Physically, this means that ½ A B U U random the same random and still has non-positive partial transposition. Thus, we can concentrate on states of the form: Z d¹ U(U - U)~ ½(U - U)y = aP+ + bP¡ 1 d ¡ where b = t r ( P ~ ) ¡½ ( x ) = P + x P We consider the (unnormalized) family of states: ½ + ¡ 2 2 = C C Qubits: H T A ½ ¸0 T A ½ ¸0 3 distillable x N y A B ) ½ ( A B ) ! j © i h © j one can easily find A, B such that ( 3 3 = C C Higher dimensions: H T A ½ ¸0 T A ½ ¸0 2 x 3 ? NPT distillable Idea: find A, B such that they project T A 2 2 ¸0 onto C with ½ -C there is a strong evidence that they are not distillable: for any finite N, all T A 2 2 ¸0 projections onto C have ½ -C What is known? PPT In general NPT ? 2xN Gaussian states (Horodecki 97, Dur et al 2000) PPT NPT (Giedke, Duan, Zoller, Cirac, 2001) PPT NPT 3. Gaussian states squeezed states: (2-mode approximation) Light source: 1 X n j v a c i = ¸ j n ; n i y y ¸ ( a b ¡ a b ) j ª i = e n = 0 Gaussian state: Decoherence: photon absorption, phase shifts ¡H ½ = e = H ( X ; P ; X ; P ) where H a a b b is at most quadratic in y y a + a a ¡ a p X = ; P = ip a a 2 2 Internal levels can be approximated Atomic ensembles: by continuous variables in Gaussian states Optical elements: - Beam splitters: - Lambda plates: - Polarizers, etc. Measurements: X, P 0 ¡ H 0 ¡ H ½ = e ! ½ = e Gaussian - Homodyne detection: Gaussian local oscillator We consider: B A m modes n modes 2 n 2 m H = [ L ( R ) ] [ L ( R ) ] ½ Gaussian Is ½ separable and/or distillable? 3.1 What is known? 1 mode + 1 mode: T A ½ is separable iff ½ = ( T 1 ) ( ½ ) ¸ 0 (Duan, Giedke, Cirac and Zoller, 2000; Simon 2000) 2 modes + 2 modes: There exist PPT entangled states. (Werner and Wolf 2000) 3.2 Separability CORRELATION MATRIX R = ( X ; P ; X ; : : : ; X ; P ; : : : ) a a a b b 1 1 2 1 1 All the information about ½ is contained in: ° = 2 R e h ( R ¡ d ) ( R̄ ¡ d̄ ) ithe „correlation matrix“. ® ® ® ; ¯ = h R i ! 0 where d ® ® µ °= 2nX2n A CT C B ¶ 2mX2m For valid density operators: °¸iJ is the „symplectic matrix“ = J © J © : : : © J where J 2 2 2 µ ¶ 0¡1 = and J 2 10 Given a CM, ° : does it correspond to a separable state (separable)? (G. Giedke, B. Kraus, M. Lewenstein, and Cirac, 2001) Idea: define a map ° ´° 0 °1 °2 °N ... T AN+ 1 = BN+ 1 = AN ¡ Re[CN (BN ¡ i J) ¡ 1CN ] CN+ 1 = ¡ Im[CN (BN ¡ i J) ¡ 1CNT ] Facts: °Nis a CM of a separable state iff ° is too. N + 1 ° N + 1is no CM If °N is a CM of an entangled state, then either or ° N + 1is a CM of an entangled state !° = ½ ½ If ° is separable, then ° . This last corresponds to ½ N 1 1 A B (for which one can readily see that is separable) CONNECTION WITH POSITIVE MAPS? ! ° N N + 1 Map for CM‘s: ° ¡ H ¡ H N N + 1 = e ! ½ = e Map for density operators: ½ N N + 1 (½N+ 1)AB = trBB~f [(½N )AB - (½N )A~B~]XB B~g ½ A A~ ½ B B~ Non-linear density operators Gaussian separable A A~ 3.3 Distillability (Giedke, Duan, Zoller, and Cirac, 2001) T ¸0 Idea: take ½ such that ½ Two modes: N=M=1: = ° Symmetric states:° A B B A T ½ s¸ 0 A A T ½ ~ s¸ 0 Non-symmetric states: B A B A distillable state. T ½ ¸0 T ½ s¸ 0 B B symmetric state. General case: N,M A T ½ N ; M̧0 B A T ½ ¸0 1 ; 1 two modes T ¸0 ½ is distillable if and only if ½ There are no NPT Gaussian states. B 4. Multipartite case. C A B ½ Are these systems entangled? Fully separable states are those that can be prepared by LOCC out of a product state. ½= N X pk jak i A hakj - jbki B hbkj - jck i Chckj k= 1 We can also consider partitions: Separable A-(BC) Separable B-(AC) C A Separable C-(AB) C B A C B A B N X p j a i h a j j 'k i h 'k j k k A k B C N X p j b i h b j j 'k i h 'k j k k B k A C N X p j c i h c j j 'k i h 'k j k k C k A B k = 1 k = 1 k = 1 4.1 Bound entangled states. Consider C A C N X ½= A B pk jak i Ahakj - j' k i BCh' k j = k= 1 N X B pk jbki B hbkj - j' k i ACh' k j k= 1 but such that it is not separable C-(AB). QUESTIONS: Is B entangled with A or C? Is A entangled with B or C? Is C entangled with A or B? Consequence: Nothing can be distilled out of it. It is a bound entangled state. 4.2 Activation of BES. (Dür and Cirac, 1999) Consider C C A A B but A and B can act jointly C A B singlets Then they may be able to distill GHZ states. B ACTIVATION OF BOUND ENTANGLED STATES Distillable iff two groups 3 and 5 particles Distillable iff two groups 35-45% and 65-55% Distillable iff two groups have more than 2 particles. 6 3 1 4 7 2 Not distillable 8 5 Not-distillable 6 4 6 3 1 7 4 2 3 1 2 Distillable 7 8 8 5 Distillable 4 2 5 6 3 1 Not-distillable 7 10 9 11 12 8 5 Not-distillable 6 3 1 4 2 Not-distillable 7 8 5 If two particles remain separated not distillable. Two parties can distill iff the other join 3 1 2 3 6 5 10 4 7 9 8 11 (Shor and Smolin, 2000) 1 2 6 5 10 4 7 1 3 12 3 C 1 9 8 4 2 12 Superactivation 4 2 A B 11 Two copies 4.3 Family of states Define: where N ¡1 There are 2 parameters. Any state can be depolarized to this form. 5. Conclusions The separability problem is one of the most challanging problems in quantum Information theory. It is relevant from the theoretical and experimental point of view. Gaussian states: Solved the separability and distillability problem for two systems. Solved the separability problem for three (1-mode) systems Maybe we can use the methods developed here to attack the general problem. Multipartite systems: New behavior regarding separability and bound entanglement. Family of states which display new activation properties. Innsbruck: Geza Giedke Hannover M. Lewenstein Barbara Kraus Wolfgang Dür Guifré Vidal J.I.C. Collaborations: R. Tarrach (Barcelona) P. Horodecki (Gdansk) L.M. Duan (Innsbruck) P. Zoller (Innsbruck) SFB Coherent Control €U TMR EQUIP KIAS, November 2001 Institute for Theoretical Physics Postdocs: - L.M. Duan (*) - P. Fedichev - D. Jaksch - C. Menotti (*) - B. Paredes - G. Vidal - T. Calarco Ph D: - W. Dur (*) - G. Giedke (*) - B. Kraus - K. Schulze P. Zoller J. I. Cirac FWF SFB F015: „Control and Measurement of Coherent Quantum Systems“ € EU networks: „Coherent Matter Waves“, „Quantum Information“ EU (IST): „EQUIP“ Austrian Industry: Institute for Quantum Information Ges.m.b.H.