Recently, the Quantum Computing Cloud Platform Team, the Superconducting Quantum Computing Team, and the Quantum Operating System Software Development Team from the Beijing Academy of Quantum Information Sciences (referred to as "BAQIS"), in collaboration with the Institute of Physics of the Chinese Academy of Sciences and the University of Science and Technology of China, have successfully demonstrated Maxwell demon-assisted Einstein-Podolsky-Rosen (EPR) steering using a 2D 9-qubit superconducting quantum chip. This achievement, titled "Demonstration of Maxwell demon-assisted Einstein-Podolsky-Rosen steering via superconducting quantum processor" was published as a Letter in Physical Review Research on September 26, 2024.

Quantum Nonlocality and EPR Steering

Nonlocality is a key feature of quantum theory that differentiates it from classical theories. The Bell inequality remains the most stringent method for verifying quantum nonlocality, a discovery recognized with the 2022 Nobel Prize in Physics. EPR steering represents a form of nonlocality that lies between Bell nonlocality and inseparability, invalidating any local hidden state model. From an information-theoretic perspective, the entropy of an entangled system after measurement is lower than that of a mixed state, implying that quantum systems contain additional information that can be harnessed to perform work. Given the profound connections between Maxwell's demon and thermodynamics as well as information theory, research into the relationship between Maxwell's demon and quantum correlations holds significant academic value.

Introducing Maxwell Demon-Assisted EPR Steering

With the rapid progress in quantum information science, there is growing interest in exploring the connection between Maxwell's demon and quantum correlations. To date, most research has focused on extracting work from quantum correlations using Maxwell's demon. Recently, BAQIS and its collaborators introduced the novel concept of Maxwell demon-assisted EPR steering (Figure 1). Their research demonstrated that simulating nonlocality in EPR steering through work is feasible, revealing a previously unnoticed thermodynamic loophole in practical quantum steering tasks.

Experimental Validation

The teams further experimentally validated their theoretical predictions. Using a superconducting quantum chip, they implemented the quantum circuit model of Maxwell demon-assisted EPR steering (Figure 2a) and observed a linear relationship between the work done by the demon and the nonlocality exhibited in EPR steering tasks (Figure 3). The superconducting quantum chip used in the study consists of a 9-qubit 2D square lattice with 9 qubits and 12 couplers (Figure 2c), with 5 qubits (q0-q4) used for executing the quantum circuit. The circuits in Figures 2a and 2b were compiled, decomposed, and optimized into sequences of single-qubit rotation gates and two-qubit CZ gates executable on the processor.

The experiment achieved average fidelities of 99.9% for single-qubit gates and 99% for two-qubit CZ gates via randomized benchmarking. The proposed Maxwell demon-assisted EPR steering scheme and its corresponding quantum circuit model demonstrate a previously unrecognized thermodynamic loophole in testing EPR steering nonlocality. This work significantly deepens our understanding of the interplay between quantum correlations, thermodynamics, and information theory, opening new directions for future exploration.

Figure 1: Illustration of Maxwell demon-assisted EPR steering.

Figure 2: (a) Quantum circuit representation of Maxwell demon-assisted EPR steering; (b) Quantum circuit without Maxwell's demon; (c) Schematic of the superconducting quantum processor.

Figure 3: (a) Experimental results verify the linear relationship between parameter p and S2, with a measured EPR steering parameter S2 of 0.770 ± 0.005, exceeding the classical boundary by 12.6 standard deviations. (b) Simulations indicate that deviations between experimental and ideal cases are primarily caused by depolarizing noise in the actual circuits.

Acknowledgments and Contributions

The first authors are Dr. Wang Ziting and Assistant Research Scientist Wang Ruixia from BAQIS. Corresponding authors are Associate Research Scientist Hu Mengjun, Assistant Research Scientist Huang Kaixuan (both from BAQIS), and Professor Zhang Yongsheng from the University of Science and Technology of China. Other contributors include Research Scientists Yu Haifeng and Zhao Shiping, Dr. Zhao Peng, and dual-affiliated Research Scientist Fan Heng, Xu Kai, Dr. Shi Yunhao, and Ph.D. candidate Yang Zhaohua. The project was supported by the National Natural Science Foundation of China, the Innovation Program for Quantum Science and Technology, and the Beijing Nova Program.

Original Source: https://doi.org/10.1103/PhysRevResearch.6.L032073