Thermodynamics of Quantum Information Flows
We report two results complementing the second law of thermodynamics for Markovian open quantum systems coupled to multiple reservoirs with different temperatures and chemical potentials. First, we derive a nonequilibrium free energy inequality providing an upper bound for a maximum power output, which for systems with inhomogeneous temperature is not equivalent to the Clausius inequality. Secondly, we derive local Clausius and free energy inequalities for the subsystems of a composite system. These inequalities differ from the total system one by the presence of an information-related contribution and build the ground for thermodynamics of quantum information processing. Our theory is used to study an autonomous Maxwell demon.
The second law of thermodynamics is one of the main principles of physics. Within equilibrium thermodynamics there exist two equivalent formulations of this law. The first, referred to as the Clausius inequality, states that the sum of entropy change of the system and the entropy exchanged with the environment during the transition between two equilibrium states is nonnegative: . The exchanged entropy can be further expressed as , where is the heat delivered to the system. An alternative formulation, referred to as the free energy inequality, states that during the transition between two equilibrium states , where is the work performed on the system and is the free energy (here denotes the internal energy). The latter formulation can be obtained from the former by using the first law of thermodynamics .
The last few decades have brought significant progress towards generalizing the second law of thermodynamics far from equilibrium to both classical [1, 2, 3, 4, 5, 6] and quantum [7, 8, 9] systems. The standard formulation generalizes the Clausius inequality by stating that the average entropy production is nonnegative. For Markovian systems the entropy production can be defined as , where is the heat delivered to the system from the reservoir with the inverse temperature [6, 9]. Formulations generalizing the free energy inequality [10, 11, 12, 13] are much less common, and have been so far confined mainly to systems coupled to an environment with a homogeneous temperature; for an exception, see Ref. .
These developments have also brought a deeper understanding of the relation between thermodynamics and the information theory [11, 14]. One of the most important achievements is related to the field of thermodynamics of feedback-controlled systems . Following the groundbreaking ideas of Maxwell demon  and Szilard engine  it was verified both theoretically [18, 19, 20, 21, 22] and experimentally [23, 24, 26, 25, 28, 27] that by employing feedback one can reduce entropy of the system without exchanging heat. In such a case modified Clausius inequalities, which relate the entropy change to the information flow, have to be applied [29, 10, 11, 34, 30, 31, 32, 33, 35, 36, 12, 38, 37, 39, 13]. It was also realized, that the feedback control does not require the presence of any intelligent being (as in the original idea of Maxwell), but may be performed by an autonomous stochastic system coupled to the controlled one . A consistent mathematical formulation of the generalized Clausius inequality for such autonomous systems operating in the classical regime has been formulated in Ref. .
Our work adds two new contributions to the field. First, we generalize the free energy inequality to Markovian open quantum systems coupled to reservoirs with different temperatures and show, that this formulation of the second law is in general not equivalent to the Clausius inequality. Secondly, we formulate a consistent thermodynamic formalism describing thermodynamics of continuous information flow in composite quantum systems and demonstrate the relation between the information and the nonequilibrium free energy. The applicability of our results is demonstrated using a quantum coherent autonomous Maxwell demon based on quantum dots.
Nonequilibrium Clausius inequality. We consider a generic open quantum system weakly coupled to equilibrated reservoirs with temperatures (inverse temperatures ) and chemical potentials , described by the time-independent Hamiltonian
where , , are, correspondingly, the Hamiltonian of the system, reservoirs and interaction of the system with the reservoirs. Within the Markov approximation the (reduced) density matrix of the system evolves according to the master equation 
where is the density matrix, denotes the total derivative of the function, is the effective Hamiltonian of the system (it may differ from due to coupling to the environment ), and is the superoperator describing the dissipative dynamics. Here and in the whole paper we take . We further assume that the dissipator is of Lindblad form, thus ensuring a completely-positive trace-preserving dynamics [42, 43]. Furthermore, within the Markov approximation the dissipation is additive, i.e., the superoperator can be represented as a sum of dissipators associated with each reservoir, denoted as : [44, 9]. For violation of additivity beyond the weak coupling regime, see Refs. [45, 46, 47, 48].
We also assume, that the grand canonical equilibrium state (Gibbs state) with respect to the reservoir
is a stationary state of , i.e., ; here is the partition function and is the particle number operator. This assumption guarantees that for an arbitrary form of dissipator the Gibbs state is a stationary state at equilibrium (i.e., for equal temperatures and chemical potentials of the reservoirs), which is true for systems weakly coupled to the environment . Let us then apply the Spohn’s inequality 
which is valid for any superoperator of Lindblad form with a steady state (not necessarily a unique steady state). As a result, one obtains the partial Clausius inequality for entropy production associated with each dissipator 
is the rate of change of the von Neumann entropy of the system due to the dissipator and
is the heat current from the reservoir . Summing all the rates one gets the total derivative of the von Neumann entropy: . Therefore, summing up Eq. (5) over the reservoirs one recovers the standard Clausius inequality 
where is the total entropy production rate. We note that the rates can be non-zero also at the steady state, when and the total entropy production is fully determined by the heat flows.
Nonequilibrium free energy inequality. Let us now define energy and work currents to the lead as
such that and , where is the internal energy. Multiplying Eq. (5) by and replacing one gets
where is the partial nonequilibrium free energy rate. Summing over one obtains the nonequilibrium free energy inequality
where is the total work rate and
is the total nonequilibrium free energy rate. Equation (12) is a complementary formulation of the second law of thermodynamics. From a practical point of view, it provides an upper bound for the maximum work output. At the steady state . and thus the system can perform work () only when a temperature difference between the reservoirs is present.
Let us emphasize, that Eqs. (8) and (12) are in general not equivalent; the former corresponds to the sum of partial second laws [Eq. (5)], whereas the latter to the weighted sum, in which Eq. (5) is multiplied by a local temperature . They become equivalent only when the system is attached to an isothermal environment, i.e., . Then the rate can be identified as the total derivative of the state function : . At the steady state and thus . This corresponds to the Kelvin-Planck statement of the second law, according to which one cannot continuously generate work by cooling an isothermal environment.
Local Clausius inequality. Let us now consider a system made of two coupled subsystems described by the Hamiltonian
where is the Hamiltonian of the subsystem and is the interaction Hamiltonian. We also assume that each subsystem is attached to a separate set of reservoirs; baths coupled with the subsystem will be then denoted as . By summing Eq. (5) over the reservoirs one obtains
Here denotes the local entropy production; it is an extensive quantity, i.e., . We will now transform Eq. (15) to a form illustrating the relation between entropy and information. Let us remind that the quantum mutual information is defined as where is the von Neumann entropy of the subsystem  (here is the reduced density matrix of the subsystem ). We can then separate the total derivative of the mutual information into two contributions: where
Here we have applied the identity . The rate can be calculated as
where is the dissipator associated with the subsystem . The rate can be further decomposed into contributions related to the unitary and the dissipative dynamics; see the Supplementary Material for details.
Replacing in Eq. (15) one obtains the local Clausius inequlity relating the entropy balance of the subsystem to the information flow:
For classical systems with the dissipator acting locally on the subsystem [i.e., for ] this inequality is equivalent to Eq. (11) from Ref.  (see the Supplementary Material). However, our result is much more general since it enables to describe quantum systems as well as classical systems with nonlocal dissipation.
Let us finally emphasize, that all the previous discussion can be easily generalized to the multipartite systems consisting of subsystems. Then , where is the multipartite mutual information .
Local free energy inequality. Analogously, summing up Eq. (11) over reservoirs we derive an inequality describing the nonequilibrium free energy balance for a single subsystem:
where and .
As in the case of the total system, Eqs. (18) and (19) are nonequivalent and complementary formulations of the local second law of thermodynamics. They become equivalent when the subsystem is coupled to an isothermal environment with a single temperature . Then and therefore
The local nonequilibrium free energy rate consists therefore of the energy-related and the information-related contribution. At the steady state the internal energy of the system is constant () and thus can be interpreted as the energy flow to the subsystem . The subsystem attached to an isothermal environment may therefore perform work either due to the energy flow from the other subsystem or due to the information flow; the latter case corresponds to the operation of information-powered devices.
Example. The applicability of our approach will now be demonstrated on a recently proposed  model of autonomous quantum Maxwell demon. Here we describe the device only briefly; for more details we refer to the original paper.
The analyzed setup [Fig. 1 (a)] is composed of two quantum dots coupled by the XY exchange interaction, each attached to two electrodes with equal temperatures . The Hamiltonian of the system is defined as
where () is the creation (annihilation) operator of an electron with spin in the dot , is the particle number operator, is the orbital energy, is the intra-dot Coulomb interaction in the dot and is the exchange coupling. The bath Hamiltonian reads , where () is the creation (annihilation) operator of an electron with spin , wave number and energy in the reservoir ; here () denotes the left (right) reservoir attached to the dot . Finally, the system bath-interaction Hamiltonian is expressed as , where is the tunnel coupling of the dot to the reservoir . We also define the coupling strength , where is the density of states of electrons with spin in the bath .
To describe the dynamics of the device we apply a microscopically derived Lindblad equation which couples populations to coherences and is thermodynamically consistent in the weak coupling limit (it is equivalent to the phenomenological approach proposed in Ref. ). Details of the method, discussion of its limits of validity and comparison with the secular Lindblad equation (which is by construction thermodynamically consistent but neglects genuine quantum coherent effects) are presented in the Supplementary Material.
The device works as follows: The baths are taken to be fully spin polarized, i.e., either or is equal to 0. As a result, the electrodes act as spin filters which forbid the tunneling of electrons with a spin opposite to the polarization [52, 53]. Additionally, the polarizations of reservoirs attached to a single dot are arranged in an anti-parallel way, such that the electron may be transferred between the electrodes only when it changes its spin. This is enabled by the XY interaction which exchanges spins between the dots [Fig. 1 (b)]. Since the XY interaction conserves the total spin, the spin flips occur simultaneously in both subsystems and thus the steady-state currents through both dots have to be equal. Let us now apply a high positive bias to the first dot, and a smaller opposite bias () to the other one. Then, the voltage-driven current through the first dot will pump electrons in the second dot against the bias, which is due to a nonequilibrium spin population induced by the spin flips. This corresponds to the operation of a Maxwell demon: the spin dynamics generated by the XY interaction (equivalent to the operation of the quantum iSWAP gate ) induces an information flow between the dots, which enables a conversion of heat into work.
Our local Clausius and free energy inequalities [Eqs. (18), (19)] for this system are demonstrated in Fig. 2. As one can see, for the dot cools its isothermal environment, which is enabled by the information flow (); at the same time it performs work (i.e, pumps the current against the voltage), which is possible due to the negative free energy rate (). This is compensated by the dissipation of work into heat in the first dot. For the energy flow is negligible () and the device is powered purely by information (). As shown in the Supplementary Material, in this regime the information current is generated by the unitary dynamics related to the coherent spin oscillations. On the other hand, for one can observe a noticeable energy flow from the first to the second dot which results from the splitting of energy levels. As a consequence, for the dot 2 starts to heat its environment () and the setup ceases to work as a Maxwell demon. However, the dot 2 still performs work thanks to the negativity of the nonequilibrium free energy rate , which now includes a significant energy-related contribution .
Conclusions. Our inequality (12), by providing a complement to the second law, may have novel implications for the design of quantum heat engines  which need to be explored. In turn, our inequalities (18) and (19) provide the basis for thermodynamics of quantum information processing. We hope that, as happened with the classical counterpart of Eq. (18), numerous applications and experiments  will also ensure it in the quantum realm.
Acknowledgements. K. P. is supported by the National Science Centre, Poland, under the project Opus 11 (No. 2016/21/B/ST3/02160) and the doctoral scholarship Etiuda 6 (No. 2018/28/T/ST3/00154). M. E. is supported by the European Research Council project NanoThermo (ERC-2015-CoG Agreement No. 681456).
-  D. J. Evans, E. G. D. Cohen, and G. P. Morriss, Probability of second law violations in shearing steady states, Phys. Rev. Lett. 71, 2401 (1993).
-  G. Gallavotti and E. G. D. Cohen, Dynamical Ensembles in Nonequilibrium Statistical Mechanics, Phys. Rev. Lett. 74, 2694 (1995).
-  J. L. Lebowitz and H. Spohn, A Gallavotti–Cohen-Type Symmetry in the Large Deviation Functional for Stochastic Dynamics, J. Stat. Phys. 95, 333 (1999).
-  U. Seifert, Entropy Production along a Stochastic Trajectory and an Integral Fluctuation Theorem, Phys. Rev. Lett. 95, 040602 (2005).
-  M. Esposito and C. Van den Broeck, Three detailed fluctuation theorems, Phys. Rev. Lett. 104, 090601 (2010).
-  U. Seifert, Stochastic Thermodynamics, Fluctuation Theorems and Molecular Machines, Rep. Prog. Phys. 75, 126001 (2012).
-  H. Spohn, Entropy Production for Quantum Dynamical Semigroups, J. Math. Phys. (N.Y.) 19, 1227 (1978).
-  M. Esposito, K. Lindenberg, and C. Van den Broeck, Entropy Production as Correlation Between System and Reservoir, New J. Phys. 12, 013013 (2010).
-  G. B. Cuetara, M. Esposito, and G. Schaller, Quantum Thermodynamics with Degenerate Eigenstate Coherences, Entropy 18, 447 (2016).
-  M. Esposito and C. Van den Broeck, Second law and Landauer principle far from equilibrium, Europhys. Lett. 95, 40004 (2011).
-  J. M. R. Parrondo, J. M. Horowitz, and T. Sagawa, Thermodynamics of information, Nat. Phys. 11, 131 (2015).
-  P. Strasberg, G. Schaller, T. Brandes, and M. Esposito, Quantum and Information Thermodynamics: A Unifying Framework Based on Repeated Interactions, Phys. Rev. X 7, 021003 (2017).
-  H. Miyahara and K. Aihara, Work relations with measurement and feedback control on nonuniform temperature systems, Phys. Rev. E 98, 042138 (2018).
-  J. Goold, M. Huber, A. Riera, L. del Rio, and P. Skrzypczyk, The role of quantum information in thermodynamics – a topical review, J. Phys. A: Math. Theor. 49, 143001 (2016).
-  K. Maruyama, F. Nori, and V. Vedral, Colloquium: The physics of Maxwell’s demon and information, Rev. Mod. Phys. 81, 1 (2009).
-  J. C. Maxwell, The Scientific Letters and Papers of James Clerk Maxwell (Cambridge University Press, Cambridge, 1995), Vol. 2.
-  L. Szilard, Über die Entropieverminderung in einem thermodynamischen System bei Eingriffen intelligenter Wesen, Z. Phys. 53, 840 (1929).
-  S. Lloyd, Quantum-mechanical Maxwellâs demon, Phys. Rev. A 56, 3374 (1997).
-  S. W. Kim, T. Sagawa, S. De Liberato, and M. Ueda, Quantum Szilard Engine, Phys. Rev. Lett. 106, 070401 (2011).
-  G. Schalller, C. Emary, G. Kiesslich, and T. Brandes, Probing the power of an electronic Maxwell’s demon: Single-electron transistor monitored by a quantum point contact, Phys. Rev. B 84, 085418 (2011).
-  D. Mandal and C. Jarzynski, Work and information processing in a solvable model of Maxwell’s demon, Proc. Nat. Acad. Sci. USA 109, 11641 (2012).
-  J. J. Park, K.-H. Kim, T. Sagawa, and S. W. Kim, Heat Engine Driven by Purely Quantum Information, Phys. Rev. Lett. 111, 230402 (2013).
-  J. V. Koski, V. F. Maisi, T. Sagawa, and J. P. Pekola, Experimental Observation of the Role of Mutual Information in the Nonequilibrium Dynamics of a Maxwell Demon, Phys. Rev. Lett. 113, 030601 (2014).
-  J. V. Koski, A. Kutvonen, I. M. Khaymovich, T. Ala-Nissila, and J. P. Pekola, On-Chip Maxwellâs Demon as an Information-Powered Refrigerator, Phys. Rev. Lett. 115, 260602 (2015).
-  P. A. Camati, J. P. S. Peterson, T. B. Batalhão, K. Micadei, A. M. Souza, R. S. Sarthour, I. S. Oliveira, and R. M. Serra, Experimental Rectification of Entropy Production by Maxwellâs Demon in a Quantum System, Phys. Rev. Lett. 117, 240502 (2016).
-  J. V. Koski and J. P. Pekola, Maxwell’s demons realized in electronic circuits, C. R. Physique 17, 1130 (2016).
-  K. Chida, S. Desai, K. Nishiguchi, and A. Fujiwara, Power generator driven by Maxwell’s demon, Nat. Commun. 8, 15310 (2017).
-  N. Cottet, S. Jezouin, L. Bretheau, P. Campagne-Ibarcq, Q. Ficheux, J. Anders, A. Auffèves, R. Azouit, P. Rouchon, and B. Huard, Observing a quantum Maxwell demon at work, Proc. Natl. Acad. Sci. USA 114, 7561 (2017).
-  T. Sagawa and M. Ueda, Generalized Jarzynski Equality under Nonequilibrium Feedback Control, Phys. Rev. Lett. 104, 090602 (2010).
-  J. M. Horowitz and S. Vaikuntanathan, Nonequilibrium detailed fluctuation theorem for repeated discrete feedback, Phys. Rev. E 82, 061120 (2010).
-  Y. Morikuni and H. Tasaki, Quantum Jarzynski-Sagawa-Ueda Relations, J. Stat. Phys. 143, 1 (2011).
-  D. Abreu and U. Seifert, Thermodynamics of Genuine Nonequilibrium States under Feedback Control, Phys. Rev. Lett. 108, 030601 (2012).
-  S. Deffner and C. Jarzynski, Information Processing and the Second Law of Thermodynamics: An Inclusive, Hamiltonian Approach, Phys. Rev. X 3, 041003 (2013).
-  T. Sagawa and M. Ueda, Role of mutual information in entropy production under information exchanges, New J. Phys. 15, 125012 (2013).
-  P. Strasberg, G. Schaller, T. Brandes, and C. Jarzynski, Second laws for an information driven current through a spin valve, Phys. Rev. E 90, 062107 (2014).
-  A. C. Barato and U. Seifert, Unifying Three Perspectives on Information Processing in Stochastic Thermodynamics, Phys. Rev. Lett. 112, 090601 (2014).
-  D. Hartich, A. C. Barato, and U. Seifert, Stochastic thermodynamics of bipartite systems: transfer entropy inequalities and a Maxwell’s demon interpretation, J. Stat. Mech. (2014) P02016.
-  A. Chapman and A. Miyake, How an autonomous quantum Maxwell demon can harness correlated information, Phys. Rev. E 92, 062125 (2015).
-  J. M. Horowitz and M. Esposito, Thermodynamics with Continuous Information Flow, Phys. Rev. X 4, 031015 (2014).
-  P. Strasberg, G. Schaller, T. Brandes, and M. Esposito, Thermodynamics of a Physical Model Implementing a Maxwell Demon, Phys. Rev. Lett. 110, 040601 (2013).
-  H.-P. Breuer and F. Petruccione, The Theory of Open Quantum Systems (Oxford University Press, Oxford, 2002).
-  G. Lindblad, On the Generators of Quantum Dynamical Semigroups, Commun. Math. Phys. 48, 119 (1976).
-  V. Gorini, A. Kossakowski, and E. C. G. Sudarshan, Completely Positive Dynamical Semigroups of N-Level Systems, J. Math. Phys. (N.Y.) 17, 821 (1976).
-  G. Schaller, Open Quantum Systems Far from Equilibrium (Springer, Heidelberg, 2014).
-  C.-K. Chan, G.-D. Lin, S. F. Yelin, and M. D. Lukin, Quantum interference between independent reservoirs in open quantum systems, Phys. Rev. A 89, 042117 (2014).
-  G. G. Giusteri, F. Recrosi, G. Schaller, and G. L. Celardo, Interplay of different environments in open quantum systems: Breakdown of the additive approximation, Phys. Rev. E 96, 012113 (2017).
-  M. T. Mitchison and M. B. Plenio, Non-additive dissipation in open quantum networks out of equilibrium, New J. Phys. 20, 033005 (2018).
-  H. M. Friedman, B. K. Agarwalla, and D. Segal, Quantum energy exchange and refrigeration: a full-counting statistics approach, New J. Phys. 20, 083026 (2018).
-  M. M. Wilde, Quantum Information Theory (Cambridge University Press, Cambridge, 2013).
-  S. Watanabe, Information theoretical analysis of multivariate correlation, IBM J. Res. Develop. 4, 66 (1960).
-  K. Ptaszyński, Autonomous quantum Maxwell’s demon based on two exchange-coupled quantum dots, Phys. Rev. E 97, 012116 (2018).
-  W. Rudziński and J. Barnaś, Tunnel magnetoresistance in ferromagnetic junctions: Tunneling through a single discrete level, Phys. Rev. B 64, 085318 (2001).
-  M. Braun, J. König, and J. Martinek, Theory of transport through quantum-dot spin valves in the weak-coupling regime, Phys. Rev. B 70, 195345 (2004).
-  N. Schuch and J. Siewert, Natural two-qubit gate for quantum computation using the XY interaction, Phys. Rev. A 67, 032301 (2003).
-  G. Kiršanskas, M. Franckié, and A. Wacker, Phenomenological position and energy resolving Lindblad approach to quantum kinetics, Phys. Rev. B 97, 035432 (2018).
-  G. Benenti, G. Casati, K. Saito, and R. S. Whitney, Fundamental aspects of steady-state conversion of heat to work at the nanoscale, Phys. Rep. 694, 1 (2017).