Quantum Thermal System

Simple SummaryContent extracted from patent full text and abstract with AI.

This patent describes a quantum thermal system that combines a quantum thermal machine (such as a quantum heat engine or refrigerator) with a computer agent using reinforcement learning. The quantum machine operates between two thermal baths and includes a quantum system (like a qubit) whose properties are adjusted over time. The computer agent autonomously tunes the machine's control parameters to maximize a desired long-term performance metric (such as average extracted power or cooling power) by learning from feedback on heat flows, without needing prior knowledge of the quantum machine's model.

Use CasesContent extracted from patent full text and abstract with AI.

• Optimizing the operation of quantum refrigerators for on-chip cooling in quantum computing hardware.
• Enhancing the efficiency of quantum heat engines used in energy harvesting or nanoscale devices.
• Adaptive control of quantum thermal machines in research laboratories for experimental studies.
• Deploying autonomous thermal management in superconducting circuit systems.
• Self-optimizing quantum machines for fundamental research in quantum thermodynamics.

BenefitsContent extracted from patent full text and abstract with AI.

• Provides model-free optimization, requiring no prior system knowledge for control, increasing versatility.
• Achieves higher long-term performance (power, efficiency) compared to traditional or hand-crafted control strategies.
• Can adapt to experimental imperfections, parameter drifts, or noise by continuous learning.
• Applicable to both real-world physical devices and computer simulations.
• Versatile framework—can optimize various objective functions relevant to both energy extraction and refrigeration.

Technical Classifications (CPCs)

Inventors & Applicants

Patent Abstract

The invention regards a quantum thermal system comprising a quantum thermal machine (10) and a computer agent (30). The quantum thermal machine (10) comprises at least two thermal baths (1, 3), each thermal bath (1, 3) characterized by a temperature, and a quantum system (2) coupled to the thermal baths (1, 3). The quantum thermal machine (10) is configured to perform thermodynamic cycles between the quantum system (2) and the thermal baths (1, 3), the thermodynamic cycles including heat fluxes (JH(t), JC(t)) flowing from the thermal baths (1, 3) to the quantum system (2), wherein the heat fluxes (JH(t), JC(t)) vary in time and are dependent on at least one time-dependent control parameter (u(t), d(t)). The computer agent (30) implements a reinforcement learning algorithm and is configured to vary the at least one time-dependent control parameter (u(t), d(t)) to change the heat fluxes (JH(t), JC(t)) such that a predefined long-term reward dependent on the heat fluxes (JH(t), JC(t)) is maximised. Further aspect of the invention regard a method for maximizing a long-term reward dependent on heat fluxes (JH(t), JC(t)) in thermodynamic cycles of a quantum thermal machine (10) and a computer agent (30) which is applicable without any prior knowledge of the quantum thermal machine.