Qiang Miao

Quantum simulations are traditionally confined to exploring dynamics starting from unentangled or low-entanglement states due to severe bottlenecks in protocol design, hardware performance, and classical verification. Here, we report the first experimental observation of non-equilibrium dynamics initiated directly from a many-body critical state. Using a fully-connected trapped-ion processor, we prepare the critical ground state of a transverse-field Ising model via a hardware-tailored, logarithmic-depth quantum circuit based on multi-scale entanglement renormalization. Following this initialization, we apply a deep Floquet drive that maintains emergent conformal symmetry, enabling us to benchmark the lattice dynamics against analytical results from continuum theory. In the resulting conformal heating phase, we extract a central charge consistent with the Ising universality class ($c=1/2$) from the universal decay of the Loschmidt echo and observe spatial energy localization predicted by field theory. Conversely, the non-heating phase exhibits global finite-time revivals. This work establishes a scalable and versatile framework for exploring critical quantum dynamics.

Quantum machine learning (QML) has attracted growing interest with the rapid parallel advances in large-scale classical machine learning and quantum technologies. Similar to classical machine learning, QML models also face challenges arising from the scarcity of labeled data, particularly as their scale and complexity increase. Here, we introduce self-supervised pretraining of quantum representations that reduces reliance on labeled data by learning invariances from unlabeled examples. We implement this paradigm on a programmable trapped-ion quantum computer, encoding images as quantum states. In situ contrastive pretraining on hardware yields a representation that, when fine-tuned, classifies image families with higher mean test accuracy and lower run-to-run variability than models trained from random initialization. Performance improvement is especially significant in regimes with limited labeled training data. We show that the learned invariances generalize beyond the pretraining image samples. Unlike prior work, our pipeline derives similarity from measured quantum overlaps and executes all training and classification stages on hardware. These results establish a label-efficient route to quantum representation learning, with direct relevance to quantum-native datasets and a clear path to larger classical inputs.

The investigation of strongly-correlated quantum matter is difficult due to the curse of dimensionality and intricate entanglement structures. These challenges are particularly pronounced in the vicinity of continuous quantum phase transitions, where quantum fluctuations manifest across all length scales. While quantum simulators give controlled access to a number of strongly correlated systems, the study of critical phenomena has been hampered by finite-size effects arising from diverging correlation lengths. Moreover, the experimental investigation of entanglement in many-body systems has been hindered by limitations in measurement protocols. To address these challenges, we employ the multiscale entanglement renormalization ansatz (MERA) and implement a holographic scheme for subsystem tomography on a fully-connected trapped-ion quantum computer. Our method accurately represents infinite systems and long-range correlations with few qubits, facilitating the efficient extraction of observables and entanglement properties, even at criticality. We observe a quantum phase transition with spontaneous symmetry breaking and reveal the evolution of entanglement properties across the critical point. For the first time, we demonstrate log-law scaling of subsystem entanglement entropies at criticality on a digital quantum computer. This achievement highlights the potential of MERA for the investigation of strongly-correlated many-body systems on quantum computers.

To enhance the variational quantum eigensolver (VQE), the CAFQA method can utilize classical computational capabilities to identify a better initial state than the Hartree-Fock method. Previous research has demonstrated that the initial state provided by CAFQA recovers more correlation energy than that of the Hartree-Fock method and results in faster convergence. In the present study, we advance the investigation of CAFQA by demonstrating its advantages on a high-fidelity trapped-ion quantum computer located at the Duke Quantum Center -- this is the first experimental demonstration of CAFQA-bootstrapped VQE on a TI device and on any academic quantum device. In our VQE experiment, we use LiH and BeH$_2$ as test cases to show that CAFQA achieves faster convergence and obtains lower energy values within the specified computational budget limits. To ensure the seamless execution of VQE on this academic device, we develop a novel hardware-software interface framework that supports independent software environments for both the circuit and hardware end. This mechanism facilitates the automation of VQE-type job executions as well as mitigates the impact of random hardware interruptions. This framework is versatile and can be applied to a variety of academic quantum devices beyond the trapped-ion quantum computer platform, with support for integration with customized packages.

We propose a variational quantum eigensolver (VQE) for the simulation of strongly-correlated quantum matter based on a multi-scale entanglement renormalization ansatz (MERA) and gradient-based optimization. This MERA quantum eigensolver can have substantially lower computation costs than corresponding classical algorithms. Due to its narrow causal cone, the algorithm can be implemented on noisy intermediate-scale quantum (NISQ) devices and still describe large systems. It is particularly attractive for ion-trap devices with ion-shuttling capabilities. The number of required qubits is system-size independent, and increases only to a logarithmic scaling when using quantum amplitude estimation to speed up gradient evaluations. Translation invariance can be used to make computation costs square-logarithmic in the system size and describe the thermodynamic limit. We demonstrate the approach numerically for a MERA with Trotterized disentanglers and isometries. With a few Trotter steps, one recovers the accuracy of the full MERA.