This is the academic gateway of Dr. Zhang Chang-kai, leading to all the works, including projects, articles and comments. My major devotion to physics concentrates on the theoretical and computational analysis of strongly correlated systems. I received my doctorate in physics from the Ludwig-Maximilians-Universität München in 2026 under the supervision of Prof. Jan von Delft, with a thesis on tensor network methods and artificial intelligence study of the two-dimensional Hubbard model, the archetypal model behind high-Tc superconductivity in cuprates.
What fascinates me most is how simple rules governing many interacting electrons give rise to remarkable collective phenomena, high-temperature superconductivity above all. I am particularly curious about the competing magnetic and superconducting tendencies in the doped Hubbard model and about how such correlated phases evolve with temperature. Beyond specific models, my interests extend to the tools themselves: pushing tensor network methods toward more challenging systems, and exploring how artificial intelligence, especially attention-based architectures, may learn and interpret quantum many-body physics from the data of simulations and cold-atom experiments.
Strongly correlated matter usually features collective degrees of freedom emerging from, yet bearing little resemblance to, its constituent particles, and identifying such emergent structure has relied heavily on physical insight. These works attempt to offer a systematic approach, with tensor network simulations providing faithful many-body states and an AI framework extracting the collective patterns from snapshots.
C. Zhang, J.-W. Li, D. Nikolaidou, and J. von Delft, Phys. Rev. Lett. 134, 116502 (2025).
Symmetric infinite projected entangled-pair state study of the ground state of the doped Hubbard model in the thermodynamic limit, showing how magnetic frustration suppresses stripe orders and stabilizes superconductivity.
C. Zhang and J. von Delft, Phys. Rev. B 114, 125114 (2026).
An enhanced exponential tensor renormalization group algorithm that cools the two-dimensional Hubbard model to unprecedentedly low temperatures, enabling a direct look at superconducting order and pseudogap behavior.
C. Zhang and J. von Delft, arXiv:2510.26864 (2025).
Attention-based artificial intelligence architectures that learn from snapshots of tensor network simulations of the Hubbard model, offering fresh perspectives on the quantum correlations behind Mott insulators, anomalous metals and superconductivity.
Fill out your name, your preferred email address and your questions or comments in the following boxes, click submit and I will receive your message.