Coherence and non-classicality in the multi-time statistics of a quantum Markovian process - Andrea Smirne - INFN Genova
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Three days in Quantum Mechanics Coherence and non-classicality in the multi-time statistics of a quantum Markovian process Andrea Smirne Institute for Theoretical Physics Genova, 7 June 2018
Outline • Introduction and motivation Is the presence of coherences in the quantum description of a physical phenomenon a synonym for its non-classicality? • General framework and main result Multi-time statistics due to sequential measurements on an open quantum system and precise link between quantum coherence and non-classicality • Non-Markovian multi-time statistics Simple examples showing how beyond the Markovian case there is no simple link between coherence and non-classicality • Outlook
Motivation: the role of quantum coherence Quantum biology Quantum thermodynamics Quantum coherence plays a key role in several physical phenomena: is this a synonym for their non-classicality? Resource theory of coherence
Noise assisted quantum(?) transport L X ✓ ◆ 1 L[⇢] = mn Lmn ⇢L†mn L†mn Lmn , ⇢ mn 2 Lmn = |mihn| several sources of (white) noise Plenio & Huelga, New J Phys 10, 113019 (2008) Efficiency The interplay between the Hamiltonian and the dissipative part of the dynamics guarantees high efficiency: noise assisted transport
Classical benchmark 12 Pauli (Förster) master equation: 2 72 classical hopping model 3 X 1 Ṗn (t) = ( 53 4 mn Pn (t) nm Pm (t)) 7 n 6 5 1.0 Full Lindblad equation 0.8 0.6 Pauli equation Pn (t) = hn|⇢(t)|ni P1HtL 0.4 Data ⇢mn (t) = 0 for m 6= n 0.2 0.0 in the full equation t @a.u.D 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Quantum coherences are needed in the quantum description, BUT… does it mean that this is a non-classical phenomenon??
Quantum coherence non-classicality The fact that QM predicts an event via Affirmation quantum coherence does not imply that there is no classical description of the consequent Classical Quantum • What is non-classical in quantum thermal machines ? • Quantifying coherence as a resource vs quantifying non-classicality How can we certify or irrevocably exclude the existence of an alternative classical description? Is there any link with quantum coherence?
Sequential measurements of one observable • Leggett-Garg inequalities Q dichotomic observable • Leggett-Garg type inequalities S. Huelga, T. Marshall, and E. Santos, Phys. Rev. A 52 (1995) Z. Zhou, S. Huelga, C.F. Li, G.C. Guo Phys. Rev. Lett. 115 (2015) • Witnesses of non-classicality C.-M. Li, N. Lambert, Y.-N. Chen, G.- Y. Chen and F. Nori Sc. Rep.2 (2012)
General framework • We start from the quantum description of a multi-time statistics Hilbert space ... Projective measurements of the observable X at instants t1 , t2 . . . tn Joint probability distribution Given this collection of probabilities as input, how can we unambiguously state that there is or there is not an alternative classical way to account for them?
Kolmogorov consistency conditions x1 x2 x3 Summing over all the possible intermediate values we simply obtain the joint probability referred to the initial and final values X P3 {x3 , t3 ; x2 , t2 ; x1 , t1 } = P2 {x3 , t3 ; x1 , t1 } x2 t1 t2 t3 X QX̂ n {xn , tn ; . . . xk+1 , tk+1 ; xk , tk ; xk 1 , tk 1 . . . x1 , t1 } = QX̂ n 1 {xn , tn ; . . . xk+1 , tk+1 ; xk , tk ; xk 1 , tk 1 . . . x1 , t1 } xk Kolmogorov extension theorem There exists a classical stochastic process whose joint probability distributions coincide with X • In the quantum realm it might not hold, as ⇢ 6= Px ⇢ ; role of coherences! x • For practical reasons, we speak about j-classicality to say that the consistency conditions hold up to ; non-classical means not even 2-CL
Open quantum systems • Any realistic description has to include the interaction with the environment, i.e., the system under study has to be treated as an open (quantum) system System: degrees of freedom we are interested in Environment: usually very complex, it is averaged out ⇢S (t) = trE U (t)[⇢S (t0 ) ⌦ ⇢E (t0 )]U † (t) = ⇤S (t)[⇢S (t0 )] Reduced dynamical maps: referred to HS only • Gorini-Kossakowski-Lindblad-Sudarshan master equation dX2 1 ✓ n o ◆ d † 1 ⇢S = L⇢S = i [H, ⇢S ] + j L ⇢ L j S j L†j Lj , ⇢S dt j=1 2 semigroup of CPT maps G. Lindblad Comm. Math. Phys. 48 (1976) 8t, s 0 V.Gorini, A. Kossakowski and E.C.G. Sudarshan, J. Math. Phys. 17 (1976)
Quantum regression theorem • Given an observable , the dynamical maps fix the one-time statistics and that conditioned w.r.t. t0 = 0 What can we say about the higher order probability distributions? • In general, we have to ’’go back’’ to the full unitary • Under proper conditions (essentially, if S-E correlations are negligible) Now only maps acting on the open system are involved: the reduced dynamical maps define the whole hierarchy of multi-time probabilities
Markovianity of the multi-time statistics • For a non-degenerate observable Markov condition • The whole hierarchy of probabilities can be reconstructed from the initial condition and the transition probabilities • We say that the statistics is j-Markovian if the QRT holds up to ; it is non-Markovian if not even 2-M. [In the spirit of Lindblad Comm. Math. Phys. 65 (1979)] Different from recent definitions, which are referred to the dynamics
Dynamics of quantum coherences • Definition: coherence-generating-and-detecting (CGD) dynamics The semigroup dynamics is CGD whenever there exist t, ⌧ 0 such that Total dephasing Applying dephasing at an intermediate time changes the state transformation with dephasing also at the initial and final times • Equivalent formulation h y |⇤(t) [| x ih x |] | z i ⇤h x̃ |⇤(⌧ )[| y ih z |]| x̃ i 6 0 = coherences are generated The same coherences are turned into populations
Dynamics of quantum coherences - II • Simple example, for a unitary dynamics |1i measurement of , eigenvectors ⇡ intermediate time t1 = 4 ⇡ final time t2 = 2 Coherences are generated AND turned into populations | 1i • NCGD maps ( ) are connected with the resource theory of coherence • Maps not creating coherence from incoherent states (MIO) • Coherence non-activating set: coherence, even if present, is not a resource Subsets of the set of NCGD maps; there are NCGD maps which are neither of the two =
Main result: one-to-one correspondence Let be a system’s non-degenerate observable and a j-Markovian statistics, then the statistics is jCL for any initial diagonal state if and only if the dynamics is NCGD • KEY POINT: we want to connect a property of the dynamics, (N)CGD, with a property of the whole hierarchy of probabilities, (N)Cl Markovianity is what allows us to do that ! QX̂ n (xn , tn . . . ; x1 , t1 ) S Classicality concerns the whole hierarchy ... marginals QX̂ 2 (x2 , t2 ; x1 , t1 ) S In general, this is not possible Dynamical quantities ‘’live’’ here
Main result: one-to-one correspondence Let be a system’s non-degenerate observable and a j-Markovian statistics, then the statistics is jCL for any initial diagonal state if and only if the dynamics is NCGD • KEY POINT: we want to connect a property of the dynamics, (N)CGD, with a property of the whole hierarchy of probabilities, (N)Cl Markovianity is what allows us to do that ! QX̂ n (xn , tn . . . ; x1 , t1 ) S Classicality concerns the whole hierarchy ... marginals QX̂ 2 (x2 , t2 ; x1 , t1 ) S Markovianity Dynamical quantities ‘’live’’ here
Sketch of the proof • Given a non-degenerate observable and a Lindblad dynamics NCGD time-homogeneous Chapman Kolmogorov equation! • Every classical, Markovian time-homogenous process satisfies this C-K Time-homogeneity of the statistics! Due to Lindblad and 2-M 2CL + 2M C-K means 2CL + 2M NCGD • j-M provides us with a notion of Markovianity which holds for any (classical and nonclassical) statistics! C-K + jM jCL
Leggett-Garg-type inequality • Introduced to provide an easily-detectable counterpart of the Leggett-Garg inequality S. Huelga, T. Marshall, and E. Santos, Phys. Rev. A 52 (1995) Given a dichotomic observable X and its correlation function under the assumptions of macroscopic realism and stationarity LGt inequality Measurements only at the initial and at different final times: easy to access • Stationarity was shown to be related with Markovianity and, actually, one can show that C-K implies the LGt inequality Given a Lindblad dynamics, the LGt inequality can be violated only if the dynamics is CGD: LGt inequality as a witness of quantum coherence • As a corollary of our main result Given a 2M statistics, the LGt inequality can be violated only if the hierarchy is non-classical: LGt inequality as a witness of non-classicality
Outline • • • Non-Markovian multi-time statistics •
The model • System: two-level system. Environment: continuous degree of freedom • Global unitary evolution ˆz |`i = `|`i ` = ±1 Example: polarization and momentum d.o.f. of a photon produced in SPDC • Initial product state with pure environmental state • Reduced dynamics: pure dephasing
Semigroup dynamics, non-Markovian statistics • Given a Lorentzian distribution 1.0 0.8 0.6 0.4 0.2 1 2 3 4 5 6 the reduced dynamics is the semigroup fixed by the Lindblad equation G.Lindblad, preprint, Stocholm (1980) • But the multi-time statistics is non-Markovian! L.Accardi, A.Frigerio, and J.Lewis, Publ RIMS Kyoto Univ. 18 (1982) ˆx |±i = ±1|±i Strong difference, even qualitative!
Non-classicality without quantum coherence • Beyond Markov, the hypothesis of our Theorem does not apply, in fact… • The statistics is non-classical: 2-time Kolmogorov does not hold • For the dynamics is NCGD (with respect to ˆx) |1i |1i |1i | i | i |+i | i |+i |+i | 1i | 1i | 1i Quantum coherence is not even generated!! • Properties of the 2-time probabilities cannot be inferred from the dynamics ⇢(0) = /2 2M and 2CL, but not 3M and 3CL
Quantum coherence without non-classicality • Distribution given by the combination of two Gaussians • The statistics is non-Markovian [and the dynamics non-semigroup] • There are (couples of) instants of time where the dynamics generates coherence and turns it into population, but the statistics is classical!! Quantum coherence only Amount of CGD (without info about higher orders) cannot be used as a witness of Violation of the non-classicality 2-t Kolmogorov
Summary • We derived a one-to-one correspondence between x1 x2 x3 Non-classicality of the multi-time statistics, in the ... sense of Kolmogorov t1 t2 t3 Capability of the dynamics to generate and detect quantum coherence • Crucial role of the Markovianity of the multi-time statistics (in the sense of QRT) QX̂ n (xn , tn . . . ; x1 , t1 ) S |1i |1i |1i ... | i | i |+i | i |+i |+i QX̂ 2 (x2 , t2 ; x1 , t1 ) S | 1i | 1i | 1i
Outlook • Application to specific physical systems (spin-boson model) and, in particular, to quantum thermodynamics • Extension of the analysis to classical invasive theories measurements influence the statistics Signaling in time
Acknowledgments
You can also read