Comparative analysis of non-equilibrium quantum Landauer bounds

Transcription

1 Comparative analysis of non-equilibrium quantum Landauer bounds Steve Campbell in collaboration with: Giacomo Guarnieri, Mauro Paternostro, and Bassano Vacchini To Appear July(ish) 2017

2 Landauer s Principle holds that "any logically irreversible manipulation of information, such as the erasure of a bit or the merging of two computation paths, must be accompanied by a corresponding entropy increase in non-informationbearing degrees of freedom of the information-processing apparatus or its environment". Endows information with strong evidence that it is a physical resource and not just some abstract entity* (*BUT an important caveat: Landauer only pertains to logically irreversible processes - information can be processed in perfectly reversible ways also) C. H. Bennet, Notes on Landauer s principle, reversible computation, and Maxwell s Demon, Studies in History and Philosophy of Modern Physics 34, (2003)

3 Landauer s Principle One of the most remarkable outcomes is the exorcism of Maxwell s Demon It asserts that energy cost to the Demon of erasing the one bit of knowledge (where the atom is) comes with an energy cost of E = k B T ln 2 This sets the lower bound on the dissipated heat from the system into its environment Q =Tr[HE (% E (t) % E (0))] Interestingly, Landauer s principle holds for both classical and quantum systems and has been experimentally verified

4 Landauer s Principle - The Experiments Classical Quantum A. Berut et al, Experimental verification of Landauer s principle linking information and thermodynamics Nature 483, (2012) R. Gaudenzi et al, Quantum-enhanced Landauer erasure and storage arxiv:

5 Landauer s Principle - The Experiments Classical Quantum A. Berut et al, Experimental verification of Landauer s principle linking information and thermodynamics Nature 483, (2012) R. Gaudenzi et al, Quantum-enhanced Landauer erasure and storage arxiv:

6 Why would we care? Landauer s bound has direct consequence in the context of computation, but our focus is more toward a thermodynamic viewpoint Information erasure goes hand in hand with a irreversible process, which leads us to think about thermodynamics and the second law Landauer s bound is interesting but does not ask (what we think is an important question): does the presence of quantumness change things? We will examine two alternative formulations of Landauer bounds and it may be tempting to think we are doing an apples to oranges comparison; however, our main goal is to assess when each bound is meaningful and how they depend on the details of the initial state C. H. Bennet, Notes on Landauer s principle, reversible computation, and Maxwell s Demon, Studies in History and Philosophy of Modern Physics 34, (2003)

7 Landauer s Principle - The Details Entropic Formulation Recall the heat dissipated by the system into its environment is given by Q =Tr[HE (% E (t) % E (0))] Landauer and Bennetts formulations relate this quantity to the associate change in entropy of the system and provide us with the entropic bound that will be our focus hqi S = S(% S (0)) S(% S (t)) Where the entropy is given by the von Neumann entropy D. Reeb and M. M. Wolf, An improved landauer principle with finite-size corrections, New J. Phys. 16, (2014)

8 Landauer s Principle Thermodynamic Formulation However, while the entropic bound arises due to Landauer s considerations, we can seek other paths to find alternative bounds on the dissipated heat Q =Tr[HE (% E (t) % E (0))] By considering the details of the dynamical map governing the evolution of the system a Landauer-like bound can be derived that involves only the Kraus operators and initial system state hqi B = ln Tr " X i K i % S(0)K i #! X K i K i = 1 J. Goold, M. Paternostro, and K. Modi, Nonequilibrium quantum landauer principle, Phys. Rev. Lett. 114, (2015) i Kraus operators s.t.: and % S (t) = X i K i % S (0)K i G. Guarnieri, S. Campbell, J. Goold, S. Pigeon, B. Vacchini, and M. Paternostro, Full counting statistics approach to the quantum non-equilibrium landauer bound, arxiv: (2017)

9 Let s get physical - The model % % S E = e H E 1i S 1i E H Z 0i S 0i E We consider the excitation preserving XX model % S (0) = H = J( x S x E + y S y E ) We know that the dissipated heat will necessarily be affected by different initial states - but what role does the quantum nature play? = w p 1 2

10 % S (0) = = w p 1 2 Β Q Dynamical Comparison Α Jt S 1i S 0i S 1i E 0i E For pure excited states the thermodynamic bound closely tracks the dissipated heat while the entropic bound appears largely useless H 1.0 Α 0.6 Changing the initial state we see that very different behaviours can be observed and that the entropic bound can be tighter S w 0 Β Q S w Jt Clearly the initial state plays a crucial role in dictating the performance of these bounds, on the actual dissipated heat, and therefore in information erasure

11 % S (0) = = w p 1 2 Which bound is better? 1i S H 1i E 0i S 0i E! 0 B > S S>B Choosing the point when <Q> is maximised we can determine which bound is tighter for the whole initial state parameter space Higher temperatures tend to favour the thermodynamic bound We see sharp boundaries between the relative performance of the bounds.but

12 but how do they really perform? 1i S H 1i E 0i S 0i E B hqi < 0 Q S It is important to caveat the previous plots with how tight either bound actually is Remarkably, both bounds can be negative despite a positive dissipated heat Also the dissipated heat can be negative

13 % S (0) = = w p 1 2 Clausius Law 1i S H 1i E 0i S 0i E It is easy to find that Clausius statement of the second law holds for Q 0 2 apple 1 2 [1 + tanh( )] We know both bounds can be negative and therefore fail to accurately capture the behaviour of the heat Thermodynamic =1 =0.1 Clausius Entropic

14 why do we care again? Different bounds on the dissipated heat of course will perform differently - but this isn t just apples to oranges comparison Sharp boundaries between the bounds appear, and more interestingly quantum coherence appears to play an almost negligible role Furthermore, the performance of these bounds sometimes fail to provide any meaningful information In the context of information erasure the role that quantum coherence plays is not fully understood yet. Our results indicate it is not particularly relevant

15 Many Thanks! Steve Campbell, Giacomo Guarnieri, Mauro Paternostro, and Bassano Vacchini, Something something Landauer Bounds To Appear arxiv July Giacomo Guarnieri, Steve Campbell, John Goold, Simon Pigeon, Bassano Vacchini, and Mauro Paternostro, Full counting statistics approach to the quantum non-equilibrium Landauer bound, arxiv: (2017)

16 Questions - (I have a few myself) Q1. Is it surprising given that a single quantum system can still only provide one bit of information after measurement? Do things change when the system is many-body and has some entanglement present? Q2. What about dimensionality - we focused on the simplest non-simple case, how about d-dimensional systems? CV systems? Q3. Other interaction models that aren t excitation preserving?