An Information-Theoretic Bound on Thermodynamic Efficiency and the Generalized Carnot’s Theorem

Anna Gabetti, Fabrizio Dolcini, Davide Girolami · 2026

Read An Information-Theoretic Bound on Thermodynamic Efficiency and the Generalized Carnot’s Theorem by Anna Gabetti, Fabrizio Dolcini, Davide Girolami (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-10762.txt.

Introduction

Gabetti, Dolcini, and Girolami derive an efficiency bound for thermal engines that can beat Carnot's limit. The bound depends on statistical correlations between the engine's internal state and Hamiltonian, can saturate even in finite-time cycles, and covers classical and quantum engines alike. It gives the exact maximum efficiency for engines running with multiple baths, tightening Carnot's ceiling. A quantum-dot engine coupled to fermionic baths achieves the bound even beyond the quasistatic regime, offering a design principle for realistic energy-harvesting machines.

Catalog record

CREATORAnna Gabetti, Fabrizio Dolcini, Davide Girolami
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE22765
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.10762

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

An Information-Theoretic Bound on Thermodynamic Efficiency and the Generalized Carnot’s Theorem Anna Gabetti, Fabrizio Dolcini, Davide Girolami Abstract We derive a bound on the efficiency of thermal engines that can be sharper than Carnot’s limit. It is a function of statistical correlations between the engine internal state and Hamiltonian, can be saturated even in finite-time cycles, and applies to both classical and quantum engines. Specifically, the bound establishes the exact maximal efficiency of engines operating with multiple baths, tightening the upper limit set by Carnot’s theorem. Then, we show that an engine made of a quantum dot coupled with fermionic baths can achieve the bound, even when operating beyond the quasistatic regime. The result provides a design principle for realistic energy harvesting machines. Introduction – Optimizing the performance of engines is a central challenge of modern technology, which requires maximizing work extraction, while minimizing energy dissipation. Specifically, the simplest thermal machine absorbs heat Q_{h}>0 from a hot reservoir (bath) at temperature T_{h} and converts part of it into work W>0 , while dumping the residual heat Q_{c}=W-Q_{h}<0 into a cold bath at temperature T_{c} . The second law of thermodynamics sets a general upper bound on the efficiency \eta of heat-to-work conversion in cycles operating between two thermal baths, stated by Carnot’s theorem 5 ; 6 : \displaystyle\eta:=1-\frac{|Q_{c}|}{Q_{h}}\leq\eta_{C},\qquad\eta_{C}=1-\frac{T_{c}}{T_{h}}. (1) However, its practical relevance is limited, for the bound is only reached in reversible cycles. This idealized scenario requires a quasi-static (infinite-time) heat exchange with the baths, while preserving their sharply defined temperatures. Real thermal machines instead interact for a finite time with baths through temperature gradients or with multiple baths at different temperatures (e.g., Otto and Diesel cycles). Furthermore, \eta_{C} provides no guideline for optimizing the use of the engine internal resources, since it only depends on environmental properties (i.e., bath temperatures). Here, we derive a bound on thermodynamic efficiency, applicable to both classical and quantum systems, which depends on controllable physical parameters of an engine. Specifically, it is a function of how much we know about its state, how fast such information changes, and the accessible energy levels, determined by the engine Hamiltonian. Since it depends on the engine internal dynamics, this bound inherently differs from the known refinements of Carnot’s theorem, which are limited to the two-bath scenario and identify an engine maximal efficiency exclusively in terms of bath temperatures 2 ; 13 ; 3 ; 10 ; 12 ; 40 ; 39 ; 23 ; 15 ; 16 ; 17 ; 37 ; 22 . Crucially, the bound can be much sharper than Eq. ( 1 ) for non-reversible cycles and for reversible ones involving multiple baths, while exactly matching \eta_{C} for reversible cycles involving two baths. In particular, we show that it can be saturated in realistic, finite time processes with imperfect control of the system driving. We study an engine that consists of a single-level quantum dot coupled to two fermionic baths of fixed temperatures and chemical potentials. This machine efficiency can reach the information-theoretic bound when energy levels are perfectly controllable, while offering suboptimal performance when environmental noise generates their uncontrollable (stochastic) fluctuations. The cycle is experimentally implementable with current technology 11 ; 34 ; 28 ; 24 ; 7 ; 14 ; 33 . Derivation of the bound – Consider a quantum system (the engine under scrutiny) in the state \rho(t) and with Hamiltonian H(t) that exchanges energy with its environment. The classical case is recovered by imposing [\rho(t),H(t)]=0,\,\forall\,t . In the Markovian regime, the system evolution is governed by the Lindblad equation: \displaystyle\dot{\rho}(t)=-i[\rho(t),H(t)]+{\cal L}_{t}(\rho(t)), (2) where {\cal L}_{t}(\rho(t)) describes the interactions with the environment 27 ; 4 , including heat exchanges with baths. The rate of change of the system internal energy is \dot{U}(t)=d\text{Tr}\{\rho(t)H(t)\}/dt , and the heat current is \dot{Q}(t):=\text{Tr}\left\{\dot{\rho}(t)H(t)\right\} 1 ; 41 . The efficiency of thermodynamic cycles, \int\dot{U}(t)dt=0 , is given by \displaystyle\eta=1-\frac{\int_{-}|\dot{Q}(t)|dt}{\int_{+}\dot{Q}(t)dt}, (3) where the integrals \int_{-(+)} are computed over time intervals in which heat is released (absorbed) by the system. Our first main result is: Result 1 – The maximal efficiency of a thermodynamic engine is \displaystyle\eta\leq\eta^{*}=1-\frac{\displaystyle\int_{-}-\mathcal{I}(t)\,dt}{\displaystyle\int_{+}\;\mathcal{I}(t)\,dt}, (4) where the maximal heat current \mathcal{I}(t) depends only on \rho(t) and H(t) . Proof. The heat current quantifies the linear dependence (covariance) between the evolution rate of the system state and its Hamiltonian 31 ; 18 : \displaystyle\dot{Q}(t)= \displaystyle\,\text{Cov}\left(\frac{d\log\rho(t)}{dt},H(t)\right), (5) \displaystyle\text{Cov}(X,Y)= \displaystyle\,\text{Tr}\{X\,Y\,\rho(t)\}-\text{Tr}\{X\,\rho(t)\}\text{Tr}\{Y\,\rho(t)\}. The covariance is maximal (in absolute value) for linearly dependent quantities, while it is zero when the two variables are not linearly correlated 9 , which corresponds to adiabatic transformations. In turn, the covariance between two variables can be bounded in terms of their covariances with respect to a third variable, which we choose to be \log(\rho(t)) . Then, one has: \displaystyle\dot{S}(t)= \displaystyle-\text{Tr}\left\{\dot{\rho}(t)\log\rho(t)\right\}=-\text{Cov}\left(\frac{d\log\rho(t)}{dt},\log\rho(t)\right) \displaystyle{G}(t):=\, \displaystyle\text{Cov}\left(H(t),\log\rho(t)\right). (6) The first term is the entropy rate of the system state, while G(t) reaches maximal absolute values for Gibbs states \rho(t)\propto e^{-\beta(t)\,H(t)} . Next, we construct the correlation matrix \text{corr}(V(t)) of the triad of variables V(t)=\left\{d\log\rho(t)/dt,\log\rho(t),H(t)\right\} . The consistency condition \text{det}\left(\text{corr}(V(t))\right)\geq 0 implies the inequality \dot{Q}(t)\leq\mathcal{I}(t) , where \displaystyle\mathcal{I}(t)= \displaystyle\left(R_{S}(t)R_{G}(t)+\sqrt{(1-R_{S}(t)^{2})(1-R_{G}(t)^{2})}\right) \displaystyle\text{SD}(H(t))\ \text{SD}\left(\frac{\mathrm{d}\log\rho(t)}{\mathrm{d}t}\right), \displaystyle R_{S}(t)= \displaystyle\frac{-\dot{S}(t)}{\text{SD}(d\log\rho(t)/dt)\,\text{SD}(\log\rho(t))}, \displaystyle R_{G}(t)= \displaystyle\frac{G(t)}{\text{SD}(\log\rho(t))\,\text{SD}(H(t))}.\quad Here, R_{S,G}(t)\in[-1,1] are the Pearson correlation coefficients of -\dot{S}(t),\,G(t) , and \text{SD}(X)=\sqrt{\text{Tr}\{\rho(t)X^{2}\}-(\text{Tr}\{\rho(t)X\})^{2}} is the standard deviation of X . Replacing the absorbed and released heat currents in Eq. ( 3 ) with \mathcal{I}(t) , one obtains the upper bound to efficiency in Eq.( 4 ). ∎ The result limits heat exchanges in terms of statistical correlations among (functions of) a system internal properties (state and Hamiltonian). In particular, reaching an efficiency \eta<\eta^{*} signals that the engine design is suboptimal. For example, given a certain initial state \rho(0) and a system-environment interaction governed by {\cal L}_{t} in Eq. ( 2 ), the maximal efficiency \eta^{*} can be reached by optimizing the Hamiltonian H(t) . We note that Eq. ( 4 ) is saturated when \det\text{corr}(V(t))=0,\forall\,t , which means that at least two variables out of the triad V(t) are linearly dependent, i.e., one of the correlation functions reaches its maximal absolute value. The only non-trivial solution of the equation is given by Gibbs states \rho(t)=e^{-\beta(t)\,H(t)}/Z(t),\,\forall t , where \beta(t) is an instantaneous inverse temperature of the system, and Z(t) is the partition function. Note that the engine does not need to be in equilibrium with the environment: the bound is saturated even if \beta(t)\neq 1/k_{\text{B}}T_{\text{env}} . That is, it can be achieved even in irreversible cycles. Yet, there exists a relation between the new bound and the standard formulation of the second law of thermodynamics in Eq. ( 1 ). When the engine reaches thermal equilibrium with the environment, the information content of its state explicitly depends on the environment temperature. Indeed, Eq. ( 4 ) generalizes Carnot’s result to the multiple-bath scenario, setting an informative upper limit to the efficiency of realistic thermal engines: Result 2 (Generalized Carnot’s Theorem) – For cyclic transformations of thermal machines, one has: \displaystyle\eta\leq\eta^{*}_{\cal R}=1-\frac{\overline{T}_{-}}{\overline{T}_{+}}, (7) where \overline{T}_{-} and \overline{T}_{+} are the entropy-weighted average temperatures of the environment, computed over regions of the cycle in which the engine releases and absorbs heat, respectively. The bound is saturated in reversible cycles, \eta_{{\cal R}}=\eta^{*}_{{\cal R}} . Proof. Consider two thermal machines \cal{R,X} that run a reversible cycle and a generic

Read the complete text