Thermalization time bounds for Pauli stabilizer Hamiltonians

Transcription

1 Thermalization time bounds for Pauli stabilizer Hamiltonians Kristan Temme California Institute of Technology arxiv: QEC 2014, Zurich

2

3 t mix apple O(N 2 e 2 )

4 Overview Motivation & previous results Mixing and thermalization The spectral gap bound Proof sketch

5 Thermalization in Kitaev s 2D model Spectral gap bound for the 2D toric code and 1D Ising R. Alicki, M. Fannes, M. Horodecki J. Phys. A: Math. Theor. 42 (2009)

6 Thermalization in Kitaev s 2D model Spectral gap bound for the 2D toric code and 1D Ising R. Alicki, M. Fannes, M. Horodecki J. Phys. A: Math. Theor. 42 (2009) Spectral gap bound: 1 3 e 8 J

7 Thermalization in Kitaev s 2D model Spectral gap bound for the 2D toric code and 1D Ising R. Alicki, M. Fannes, M. Horodecki J. Phys. A: Math. Theor. 42 (2009) c 1 Sn d 1 Co Spectral gap bound: 1 3 e 8 J Implies mixing time bound: c 2 d 2

8 Thermalization in Kitaev s 2D model Spectral gap bound for the 2D toric code and 1D Ising R. Alicki, M. Fannes, M. Horodecki J. Phys. A: Math. Theor. 42 (2009) c 1 Sn d 1 Co Spectral gap bound: 1 3 e 8 J Implies mixing time bound: t mix apple O(Ne 8 J ) c 2 d 2

9 The energy barrier 0 i 1 i

10 The energy barrier Arrhenius law t mem e E B Phenomenological law of the lifetime Bravyi, Sergey, and Barbara Terhal, J. Phys. 11 (2009) Olivier Landon-Cardinal, David Poulin Phys. Rev. Lett. 110, (2013) 0 i 1 i

11 The energy barrier Arrhenius law t mem e E B Phenomenological law of the lifetime Bravyi, Sergey, and Barbara Terhal, J. Phys. 11 (2009) Olivier Landon-Cardinal, David Poulin Phys. Rev. Lett. 110, (2013) 0 i 1 i Question: Can we prove a connection between the energy barrier and thermalization?

12 Stabilizer Hamiltonians Example : Toric Code A set of commuting Pauli matrices Kitaev, A. Y. (2003). Fault-tolerant quantum computation by anyons. Annals of Physics, 303(1), 2 30.

13 Stabilizer Hamiltonians Example : Toric Code A set of commuting Pauli matrices The Stabilizer Group Logical operators Kitaev, A. Y. (2003). Fault-tolerant quantum computation by anyons. Annals of Physics, 303(1), 2 30.

14 Stabilizer Hamiltonians Example : Toric Code A set of commuting Pauli matrices The Stabilizer Group Logical operators Stabilizer Hamiltonian Kitaev, A. Y. (2003). Fault-tolerant quantum computation by anyons. Annals of Physics, 303(1), 2 30.

15 Open system dynamics Lindblad master equation = L( ) = i[h, ]+ X k L k L k 1 2 {L k L k, } +

16 Open system dynamics Lindblad master equation = L( ) = i[h, ]+ X k L k L k 1 2 {L k L k, } +

17 Open system dynamics Lindblad master equation = L( ) = i[h, ]+ X k L k L k 1 2 {L k L k, } + With a unique fixed point

18 Thermal noise model & Weak coupling limit

19 Thermal noise model & Weak coupling limit

20 Thermal noise model & Weak coupling limit The evolution : S (t + t) =tr R [e ih t ( (t) R )e ih t ]

21 Thermal noise model & Weak coupling limit The evolution : S (t + t) =tr R [e ih t ( (t) R )e ih t ] Weak coupling limit & Markovian approximation: Davies, E. B. (1974). Markovian master equations. Communications in Mathematical Physics, 39(2),

22 The Davies generator

23 The Davies generator For a single thermal bath: KMS conditions*: Ensures detail balance with: Gibbs state as steady state S (!) =e! S (!) / e H S * Kubo, R. (1957). Statistical-Mechanical Theory of Irreversible Processes. I. General Theory and Simple Applications to Magnetic and Conduction Problems. Journal of the Physical Society of Japan, 12(6), Martin, P., & Schwinger, J. (1959). Theory of Many-Particle Systems. I. Physical Review, 115(6),

24 Davies generator for Pauli stabilizers Lindblad operators e iht S e iht = X! S (!)e i!t

25 Davies generator for Pauli stabilizers Lindblad operators e iht S e iht = X! S (!)e i!t Syndrome projectors S (!) = X i P (a)!= a a

26 Davies generator for Pauli stabilizers Lindblad operators e iht S e iht = X! S (!)e i!t Syndrome projectors S (!) = X i P (a)!= a a The Lindblad operators are local! (when the code is)

27 Davies generator for Pauli stabilizers Lindblad operators e iht S e iht = X! S (!)e i!t Syndrome projectors S (!) = X i P (a)!= a a The Lindblad operators are local! (when the code is)

28 Convergence to the fixed point σ For a unique fixed point: t>t mix ( ) ) ke Lt ( 0 ) k tr apple

29 Convergence to the fixed point σ For a unique fixed point: t>t mix ( ) ) ke Lt ( 0 ) k tr apple Exponential convergence ke tl ( 0 ) k tr apple Ae Bt

30 Convergence to the fixed point σ For a unique fixed point: t>t mix ( ) ) ke Lt ( 0 ) k tr apple Exponential convergence Temme, K., et al. "The χ2-divergence and mixing times of quantum Markov processes." Journal of Mathematical Physics (2010):

31 Convergence to the fixed point σ For a unique fixed point: t>t mix ( ) ) ke Lt ( 0 ) k tr apple Exponential convergence Temme, K., et al. "The χ2-divergence and mixing times of quantum Markov processes." Journal of Mathematical Physics (2010): A thermal σ implies the bound k 1 k e c N =) t mix O( N 1 )

32 Spectral gap bound Theorem 14 For any commuting Pauli Hamiltonian H, eqn. (1),the spectral gap λ of the Davies generator L β,c.f.eqn(15),withweightonepaulicouplingsw 1 is bounded by λ h exp( 2β ϵ), (81) 4η O

33 Spectral gap bound Theorem 14 For any commuting Pauli Hamiltonian H, eqn. (1),the spectral gap λ of the Davies generator L β,c.f.eqn(15),withweightonepaulicouplingsw 1 is bounded by λ h exp( 2β ϵ), (81) 4η O The constants are: The largest Pauli path: = O(N) smallest transition rate: generalized energy barrier :

34 Generalized energy barrier Paths on the Pauli Group

35 Generalized energy barrier Paths on the Pauli Group

36 Generalized energy barrier Paths on the Pauli Group

37 Generalized energy barrier Paths on the Pauli Group

38 Generalized energy barrier Paths on the Pauli Group Reduced set of generators

39 Generalized energy barrier Paths on the Pauli Group Reduced set of generators Energy barrier of the Pauli

40 Generalized energy barrier Paths on the Pauli Group Reduced set of generators Energy barrier of the Pauli

41 Generalized energy barrier Paths on the Pauli Group Reduced set of generators Energy barrier of the Pauli The generalized energy barrier

42 Generalized energy barrier Paths on the Pauli Group Reduced set of generators Energy barrier of the Pauli The generalized energy barrier Example: 2D Toric Code

43 3D Toric Code Consider the toric code on an lattice H = J X v A v J X p B p

44 3D Toric Code Consider the toric code on an lattice

45 3D Toric Code Consider the toric code on an lattice leads to a bound

46 Fernando Pastawski 3D Toric Code Michael Kastoryano Consider the toric code on an lattice leads to a bound High temperature bound

47 Discussion of the bound Relationship to Arrhenius law t mem e E B

48 Discussion of the bound Relationship to Arrhenius law t mem e E B It would be nicer to have a bound that includes entropic contributions

49 Discussion of the bound Relationship to Arrhenius law t mem e E B It would be nicer to have a bound that includes entropic contributions Can we get rid of the 1/N factor?

50 Proof sketch The Poincare Inequality Matrix pencils and the PI The canonical paths bound The spectral gap and the energy barrier

51 The Poincare Inequality Var (f,f) apple E(f,f)

52 The Poincare Inequality tr f f tr [ f] 2 apple tr f L(f)

53 The Poincare Inequality tr f f tr [ f] 2 apple tr f L(f) For classical Markov processes Sampling the Permanent : M. Jerrum, A. Sinclair. "Approximating the permanent." SIAM journal on computing 18.6 (1989): Powerful because it can lead to a geometric interpretation Cheeger s bound Canonical paths

54 The Poincare Inequality tr f f tr [ f] 2 apple tr f L(f) For classical Markov processes Sampling the Permanent : M. Jerrum, A. Sinclair. "Approximating the permanent." SIAM journal on computing 18.6 (1989): Powerful because it can lead to a geometric interpretation Cheeger s bound Canonical paths Challenges in the quantum setting We are missing a general geometric picture

55 Poincare and a Matrix pencil 1 = Equivalent formulation for Var (f,f) apple E(f,f) minimize subject to Ê ˆV 0 where E(f,f) =(f Ê f) and Var (f,f) =(f ˆV f)

56 Poincare and a Matrix pencil 1 = Equivalent formulation for Var (f,f) apple E(f,f) minimize subject to Ê ˆV 0 where E(f,f) =(f Ê f) and Var (f,f) =(f ˆV f) Lemma: Let and Ê = AA ˆV = BB =minkw k 2 subject to AW = B Boman, Erik G., and Bruce Hendrickson. "Support theory for preconditioning." SIAM Journal on Matrix Analysis and Applications 25.3 (2003):

57 Suitable matrix factorization Some intuition from E(f,f)! 0 L(f) X i: i ( i i f i i f)

58 Suitable matrix factorization Some intuition from E(f,f) Var(f,f)! 0 L(f) X i: i ( i i f i i f) V(f) 1 4 N X ( N N f N N f)

59 Suitable matrix factorization Some intuition from E(f,f) Var(f,f)! 0 L(f) X i: i ( i i f i i f) V(f) 1 4 N X ( N N f N N f) Choosing a decomposition in terms of ( x 1 f x 1 f)+( z 2f z 2 f)+( x 3 f x 3 f) ( x 1 z 2 x 3 f x 1 z 2 x 3 f)

60 Suitable matrix factorization Some intuition from E(f,f) Var(f,f)! 0 L(f) X i: i ( i i f i i f) V(f) 1 4 N X ( N N f N N f) Choosing a decomposition in terms of ( x 1 z 2 f x 1 z 2 f)+( x 3 f x 3 f) ( x 1 z 2 x 3 f x 1 z 2 x 3 f)

61 Suitable matrix factorization Some intuition from E(f,f) Var(f,f)! 0 L(f) X i: i ( i i f i i f) V(f) 1 4 N X ( N N f N N f) Choosing a decomposition in terms of ( x 1 z 2 f x 1 z 2 f)+( x 3 f x 3 f) ( x 1 z 2 x 3 f x 1 z 2 x 3 f) A generalization yields to the matrix triple [A, B, W ] kw k 2 can be bounded by suitable norm bounds

62 Canonical paths bound The norm bound on kw k 2 can be evaluated in the following picture Dressed Pauli paths :

63 Canonical paths bound The norm bound on kw k 2 can be evaluated in the following picture Dressed Pauli paths :

64 Canonical paths bound The norm bound on kw k 2 can be evaluated in the following picture Dressed Pauli paths :

65 Canonical paths bound The norm bound on kw k 2 can be evaluated in the following picture Dressed Pauli paths : The matrix norm bound yields apple max 4 X 2 N h(! (b)) b ˆ a 2 ( ) a a

66 The spectral gap and the energy barrier The only challenge is the maximum in the definition of

67 The spectral gap and the energy barrier The only challenge is the maximum in the definition of Injective map (Jerrum & Sinclair) : ( )! P N

68 The spectral gap and the energy barrier The only challenge is the maximum in the definition of Injective map (Jerrum & Sinclair) : ( )! P N [ (ˆ a )] k = (0, 0)k : k apple : k> k

69 The spectral gap and the energy barrier The only challenge is the maximum in the definition of Injective map (Jerrum & Sinclair) : ( )! P N [ (ˆ a )] k = (0, 0)k : k apple : k> k Bounding b + b b b apple 2

70 The spectral gap and the energy barrier The only challenge is the maximum in the definition of Injective map (Jerrum & Sinclair) : ( )! P N [ (ˆ a )] k = (0, 0)k : k apple : k> k Bounding h (! (a)) a b (ˆ b ) h e 2 b b

71 The spectral gap and the energy barrier The only challenge is the maximum in the definition of Injective map (Jerrum & Sinclair) : ( )! P N [ (ˆ a )] k = (0, 0)k : k apple : k> k Bounding h (! (a)) a b (ˆ b ) h e 2 b b 0 apple 4 h e 2 max ˆ X ˆ b 2 (ˆ ) 1 2 N b (ˆ b )

72 The spectral gap and the energy barrier The only challenge is the maximum in the definition of Injective map (Jerrum & Sinclair) : ( )! P N [ (ˆ a )] k = (0, 0)k : k apple : k> k Bounding h (! (a)) a b (ˆ b ) h e 2 b b 0 apple 4 h e 2 max ˆ X ˆ b 2 (ˆ ) 1 2 N b (ˆ b ) {z } apple 1

73 Conclusion and Open Questions Is it possible to find a bound that also takes the entropic contributions into account? Can we get rid of the prefactor? It would be great if one could extend the results to more general quantum memory models. This only provides a converse to the lifetime of the classical memory. It would be great if one could find a converse for the quantum memory time

74