Embedded Files
Quantum scrambling dynamics in quantum circuit (with mid-circuit measurements)
Quantum scrambling dynamics in quantum circuit (with mid-circuit measurements)
We study a joint quantum system of data and bath. In each step, the joint system is evolved by layers of unitaries and followed by mid-circuit measurements and reset on the bath system. Using this circuit architecture, we propose the ''Holographic deep thermalization'' for hardware-efficient random state generation with security against entanglement attack [1]. Recently, we also study the quantum information loss in the circuit for both monitored and unmonitored measurements [2]. In monitored dynamics, the QMI can exist for exponentially long while for unmonitored dynamics, the QMI only persists in linear lifetime.
We also study the entanglement growth in passive Gaussian random quantum circuits, and identify a parabolic entanglement lightcone [3].
Related publications
[1] Holographic deep thermalization for secure and efficient quantum random state generation, B. Zhang, P. Xu, X. Chen, Q. Zhuang, Nat Commun 16, 6341 (2025).
[2] Scaling Laws of Quantum Information Lifetime in Monitored Quantum Dynamics, B. Zhang, F. Hu, R. Mo, T. Chen, HE Türeci, Q. Zhuang, arXiv: 2506.22755 (2025).
[3] Entanglement formation in continuous-variable random quantum networks, B. Zhang, Q. Zhuang, npj Quantum Inf. 7, 33 (2021).
Training of quantum neural networks & variational quantum algorithm
Training of quantum neural networks & variational quantum algorithm
We provide an analytical theory to describe the gradient descent-based training dynamics in quantum neural networks for optimization [1] and supervised learning [2]. We derive the dynamical equations for training and reveal a dynamical transition controlled by the target value of loss function and input quantum data. We also propose the restricted Haar random ensemble to characterize the circuit ensemble in late-time training. Our theory guides the design of loss function to speed up the training.
Besides, we connect the hardness of training in quantum approximate optimization algorithm (QAOA) to the controllability of circuits in solving combinatorial problems such as SAT [3]. We also analytically analyze the trainability of universal variational quantum circuit in continuous-variable systems for the first time, and identify the dependence on circuit energy [4].
Related publications
[1] Dynamical transition in controllable quantum neural networks with large depth, B. Zhang, J. Liu, X-C. Wu, L. Jiang, Q. Zhuang, Nat Commun 15 9354 (2024).
[2] Quantum-data-driven dynamical transition in quantum learning, B. Zhang, J. Liu, L. Jiang, Q. Zhuang, npj Quantum Inf. 11, 132 (2025).
[3] Quantum computational phase transition in combinatorial problems, B. Zhang, A. Sone, Q. Zhuang, npj Quantum Inf. 8, 87 (2022).
[4] Energy-dependent barren plateau in bosonic variational quantum circuits, B. Zhang, Q. Zhuang, Quantum Sci. Technol. 10, 015009 (2024).
Variational quantum algorithms for quantum machine learning
Variational quantum algorithms for quantum machine learning
We propose the Quantum denoising diffusion probablistic model (QuDDPM), a quantum analog to the classical diffusion model, for generative quantum machine learning, targeted at the generative learning of a quantum state ensemble [1, 2]. Besides, we also apply variational quantum algorithm to study many-body physics [3], quantum sensing [4, 5] and entanglement distillation [6].
Related publications
[1] Generative quantum machine learning via denoising diffusion probabilistic models, B. Zhang, P. Xu, X. Chen, Q. Zhuang, Phys. Rev. Lett. 132, 100602 (2024).
[2] Mixed-state quantum denoising diffusion probabilistic model, G. Kwun, B. Zhang, Q. Zhuang, Phys. Rev. A 111, 032610 (2025).
[3] Uncovering Quantum Many-body Scars with Quantum Machine Learning, J. Feng, B. Zhang, Z-C. Yang, Q. Zhuang, npj Quantum Inf. 11, 42 (2025).
[4] Fast decay of classification error in variational quantum circuits. Quantum. Sci. Technol. 7, 035017 (2022).
[5] Quantum-enhanced learning with a controllable bosonic variational sensor network, P. Liao, B. Zhang, Q. Zhuang, Quantum. Sci. Technol. 9, 045040 (2024).
[6] Hybrid entanglement distribution between remote microwave quantum computers empowered by machine learning. Phys. Rev. Appl. 18, 064016 (2022).
More interesting work coming soon!
More interesting work coming soon!
Page updated
Google Sites
Report abuse