Quantum Correlations Boost Engine Efficiency to Natural Limit

Piotr Ćwikliński and colleagues at University of Gdańsk demonstrate that the maximal average feedback work in a Szilard-type engine, influenced by correlations from a CHSH prediction task, is limited by $k_B T \ln $2 multiplied by the mutual information between the system’s microstate and the controller bit. A clear relationship between the CHSH value and reversible feedback work is revealed, offering a thermodynamic valuation of side information and clarifying that Bell nonlocality does not, in itself, represent a source of free energy. The findings rigorously order the performance of classical, quantum, and nonsignalling scenarios, providing valuable insight into the fundamental limits of information processing and thermodynamics.

Thermodynamic work limits defined by Bell correlations and mutual information

A quantified maximal average feedback work of kBT ln 2 multiplied by mutual information is now achievable through extraction using correlations derived from Bell-type predictions. This represents a sharp improvement over previous methods, which were unable to precisely define this limit. The bound establishes a clear threshold for reversible work extraction, previously unattainable without understanding the interaction between correlation strength and thermodynamic value. Historically, the connection between quantum correlations and thermodynamic work has been a subject of intense debate, with earlier studies often providing only probabilistic or context-dependent bounds. This research provides a rigorous, information-theoretic upper limit, grounded in the principles of both thermodynamics and quantum mechanics. The significance lies in establishing a fundamental constraint on how efficiently information, encoded in these correlations, can be converted into useful work. The logarithmic term, ln 2, arises from the inherent limitations imposed by the second law of thermodynamics and the nature of information processing, reflecting the maximum possible efficiency of converting information into work.

Careful analysis orders classical, quantum, and nonsignalling systems based on their potential for work, offering a new perspective on the fundamental limits of information processing. As demonstrated using a Szilard-type engine powered by a thermal system and utilising Bell-type correlations, the maximum average work obtainable from these correlations is limited by kBT ln 2 multiplied by the mutual information between the system and a controller bit. The success probability of the CHSH embedding was found to be 1/2 plus the CHSH value divided by eight, directly impacting the achievable work. This success probability is crucial as it dictates the reliability with which the controller bit can predict the microstate of the thermal system, thereby influencing the efficiency of the feedback process. The Szilard engine, in this context, operates by exploiting information about the system’s microstate to perform work on a piston, effectively converting knowledge into mechanical energy. The researchers employed a two-level system, meaning the thermal system can exist in one of two distinct energy states, simplifying the analysis while still capturing the essential physics. However, this analysis assumes an ideal scenario and does not account for the energy required to create and maintain the correlation resource itself, a significant hurdle for practical application. This reveals a clear hierarchy, with classical, quantum, and nonsignalling systems each possessing distinct ceilings for work extraction, ordered by their efficiency. Classical systems, limited by classical correlations, exhibit the lowest potential for work extraction, followed by quantum systems leveraging entanglement, and finally nonsignalling systems which, while not violating causality, can exhibit stronger correlations than either classical or quantum systems. The ordering provides a benchmark for evaluating the thermodynamic performance of different information processing paradigms.

Bell correlations and thermodynamic work extraction in a Szilard engine

Increasing attention focuses on extracting useful work from fundamental properties of information, seeking to redefine the boundaries of thermodynamics. This quantification of the thermodynamic value of correlations, specifically those arising from Bell-type predictions, occurs within a Szilard engine, a conceptual device converting information into work. The Szilard engine, originally proposed by Leó Szilárd in 1929, serves as a thought experiment illustrating the link between information and energy. In its original form, it considers a single particle in a box with a partition, and work is extracted by using information about which side of the box the particle is on. This modern adaptation incorporates the complexities of quantum correlations and nonsignalling principles, extending the engine’s capabilities beyond the classical realm. The side-information channel, or data about a system gained from another, is highlighted as important, and quantifying its value can improve understanding of these quantum phenomena and their potential contribution to future technologies. The controller bit, G, acts as this side information, providing a prediction about the microstate, X, of the thermal system. The strength of the correlation between G and X, quantified by the mutual information, directly determines the amount of work that can be extracted.

It is vital to acknowledge that establishing and validating these correlations incurs energetic costs absent from this initial modelling. Creating and maintaining the Bell-type correlations requires energy expenditure, potentially offsetting the gains from work extraction. This is a crucial consideration for any potential technological application, as the net energy balance must be positive for the system to be viable. Future research should focus on minimising these overhead costs through optimised experimental designs and resource management strategies. This provides a valuable thermodynamic baseline for assessing the potential of Bell-type correlations as a resource for work extraction, while also clarifying that Bell nonlocality, a quantum phenomenon, does not itself generate free energy. Bell nonlocality, demonstrated through violations of Bell inequalities like the CHSH inequality, signifies a fundamental departure from classical physics, but it does not imply the existence of a perpetual motion machine. The energy required to establish the nonlocality remains subject to the laws of thermodynamics. A quantifiable thermodynamic value for correlations represents a step forward in understanding information’s limits. When employed within a Szilard engine, Bell-type correlations induce side information possessing a demonstrable thermodynamic value, referring to extra data about a system that can be used to improve its performance. This improvement stems from the ability to more accurately predict the system’s state, reducing the uncertainty and increasing the efficiency of the work extraction process. The analysis carefully categorises classical, quantum, and nonsignalling systems, revealing distinctions in their potential for work extraction. This categorization is not merely academic; it has implications for the development of novel information processing technologies, potentially leading to more efficient and powerful devices.

The research demonstrates that side information generated by Bell-type correlations has a quantifiable thermodynamic value when used in a Szilard engine. This means that the extra information about a system’s state, induced by these correlations, can be leveraged to improve the efficiency of work extraction. The maximal average feedback work achieved was found to be limited by the amount of this side information, specifically related to the CHSH value and binary symmetric channel success probability. The authors suggest future work will focus on minimising the energy costs associated with creating and maintaining these correlations, to better assess their viability as a resource.