Artificial intelligence (AI) has become an exceptionally powerful tool for analyzing scientific data. In particular, attention-based architectures have demonstrated a remarkable capability to capture complex correlations and to furnish interpretable insights into latent, otherwise inconspicuous patterns. This progress motivates the application of AI techniques to the analysis of strongly correlated electrons, which remain notoriously challenging to study using conventional theoretical approaches. Here, we propose AI workflows for analyzing snapshot datasets from tensor-network simulations of the two-dimensional (2D) Hubbard model over a broad range of temperature and doping. The 2D Hubbard model is an archetypal strongly correlated system, hosting diverse intriguing phenomena including Mott insulators, anomalous metals, and high- T c superconductivity. Our AI techniques yield fresh perspectives on the intricate quantum correlations underpinning these phenomena and facilitate universal omnimetry for ultracold-atom simulations of the corresponding strongly correlated systems.
Artificial intelligence (AI) has become an exceptionally powerful tool for analyzing scientific data. In particular, attention-based architectures have demonstrated a remarkable capability to capture complex correlations and to furnish interpretable insights into latent, otherwise inconspicuous patterns. This progress motivates the application of AI techniques to the analysis of strongly correlated electrons, which remain notoriously challenging to study using conventional theoretical approaches. Here, we propose AI workflows for analyzing snapshot datasets from tensor-network simulations of the two-dimensional (2D) Hubbard model over a broad range of temperature and doping. The 2D Hubbard model is an archetypal strongly correlated system, hosting diverse intriguing phenomena including Mott insulators, anomalous metals, and high- T c superconductivity. Our AI techniques yield fresh perspectives on the intricate quantum correlations underpinning these phenomena and facilitate universal omnimetry for ultracold-atom simulations of the corresponding strongly correlated systems.
The two-dimensional (2D) Hubbard model, widely believed to capture the essential physics of high-T_c cuprate compounds, has attracted immense research interest ever since the discovery of superconductivity in these materials. Despite its superficial simplicity, the Hubbard model incubates abundant fascinating phenomena owing to the strong interactions between charge carriers, encompassing antiferromagnetism, the pseudogap, and unconventional superconductivity. Yet these very interactions induce strong correlations, rendering canonical perturbative approaches unreliable. With advances in modern computational hardware, numerical techniques — notably tensor network methods and Artificial Intelligence (AI) algorithms — have become indispensable to our expedition. Tensor networks encode the amplitude information of many-body quantum states in a network of interconnected tensors. Leveraging the entanglement area law, tensor networks provide efficient, systematic, and controllable representations of quantum states that would otherwise require an exponentially large number of parameters. Past decades have witnessed the development of various tensor network ansatzes, such as Matrix Product States/Operators (MPS/MPO) for one-dimensional systems and Projected Entangled-Pair States (PEPS) for two-dimensional systems, together with diverse algorithms for optimizing these ansatzes for both zero- and finite temperature scenarios. Over the past years, AI technologies have revolutionized the way we analyze scientific data. The transformer architecture, in particular, has demonstrated exceptional capabilities in the domain of natural language processing. Its attention mechanism excels at capturing long-range correlations in sequential data, making it a promising tool for studying strongly correlated many-body systems. The interpretable nature of the attention mechanism further offers valuable insights into the AI's perception of the underlying physics. In this thesis, we assemble an all-inclusive numerical toolchain for analyzing strongly correlated lattice systems. As a demonstration, we investigate the zero- and finite-temperature properties of the 2D Hubbard model on a square lattice via the infinite PEPS (iPEPS) and the eXponential Tensor Renormalization Group (XTRG), respectively. The native two-dimensional iPEPS tensor network faithfully embodies the entanglement area law and helps settle long-standing debates regarding superconducting order in the ground state of the Hubbard model. The XTRG algorithm efficiently constructs thermal density matrices across a broad temperature range via a highly effective exponential-cooling protocol, thereby enabling detailed probes of pairing and pseudogap phenomena. Afterwards, we generate a comprehensive snapshot dataset which furnishes subsequent training of an optimized encoder-only transformer model. Our novel AI architecture features a Markovian interpretation of the attention design and affords improved parallelism. The collective technological stack opens a brand new avenue for exploring the rich physics of strongly correlated many-body systems.
Strongly correlated phases of matter are often described in terms of straightforward electronic patterns. This has so far been the basis for studying the Fermi-Hubbard model realized with ultracold atoms. Here, we show that artificial intelligence (AI) can provide an unbiased alternative to this paradigm for phases with subtle, or even unknown, patterns. Long- and short-range spin correlations spontaneously emerge in filters of a convolutional neural network trained on snapshots of single atomic species. In the less well-understood strange metallic phase of the model, we find that a more complex network trained on snapshots of local moments produces an effective order parameter for the non-Fermi-liquid behavior. Our technique can be employed to characterize correlations unique to other phases with no obvious order parameters or signatures in projective measurements, and has implications for science discovery through AI beyond strongly correlated systems.
Image-like data from quantum systems promises to offer greater insight into the physics of correlated quantum matter. However, the traditional framework of condensed matter physics lacks principled approaches for analyzing such data. Machine learning models are a powerful theoretical tool for analyzing image-like data including many-body snapshots from quantum simulators. Recently, they have successfully distinguished between simulated snapshots that are indistinguishable from one and two point correlation functions. Thus far, the complexity of these models has inhibited new physical insights from such approaches. Here, we develop a set of nonlinearities for use in a neural network architecture that discovers features in the data which are directly interpretable in terms of physical observables. Applied to simulated snapshots produced by two candidate theories approximating the doped Fermi-Hubbard model, we uncover that the key distinguishing features are fourth-order spin-charge correlators. Our approach lends itself well to the construction of simple, versatile, end-to-end interpretable architectures, thus paving the way for new physical insights from machine learning studies of experimental and numerical data.
Strongly correlated materials exhibit complex emergent phenomena -- such as unconventional high-temperature superconductivity -- that cannot be explained within a single or weakly-interacting particle description, making understanding their physics a central challenge in condensed matter physics. This thesis develops and applies novel approaches to strongly correlated electron systems, focusing on neural quantum states (NQS) and quantum simulation with cold atoms in optical lattices or Rydberg tweezer arrays. A key contribution of this work is the advancement of neural-network-based representations of quantum many-body states. By employing state-of-the-art network architectures like Transformers and combining them with physical constructions such as determinants, Pfaffians and Gutzwiller-projection, NQS are extended to simulate large two-dimensional fermionic single- and multi-band systems with high accuracy and beyond the capabilities of conventional methods. In addition, the thesis explores how quantum simulators based on ultracold atoms in optical lattices can be leveraged to study models directly relevant to strongly correlated materials -- e.g. the multi-band model of cuprate superconductors -- as well as proposes strategies to probe superconductivity in this setting. Furthermore, hybrid approaches that combine NQS with data from quantum simulators are developed. This synergy allows for improved optimization and direct comparison between theory and experiment. These methodological developments are applied to investigate two fundamental questions in high-temperature superconductivity: On the one hand, the work proposes a microscopic mechanism based on Feshbach resonances and tests this hypothesis extensively in a model system relevant to cuprates, nickelates and heavy fermion materials. On the other hand, it develops systematic approaches to derive effective low-energy models from more complex systems like multi-band models or electron-phonon coupled systems, shining light on the timely question which single-band models capture the physics of high-temperature superconductors like cuprate or nickelate materials. Overall, the thesis demonstrates how combining machine learning and quantum simulation provides powerful tools for understanding strongly correlated quantum matter.
Abstract The rapid development of neural quantum states (NQS) has established it as a promising framework for studying quantum many-body systems. In this work, by leveraging the cutting-edge transformer-based architectures and developing highly efficient optimization algorithms, we achieve the state-of-the-art results for the doped two-dimensional (2D) Hubbard model, arguably the minimum model for high-Tc superconductivity. Interestingly, we find different attention heads in the NQS ansatz can directly encode correlations at different scales, making it capable of capturing long-range correlations and entanglements in strongly correlated systems. With these advances, we establish the half-filled stripe in the ground state of 2D Hubbard model with the next nearest neighboring hopping, consistent with experimental observations in cuprates. Our work establishes NQS as a powerful tool for solving challenging many-fermions systems.
Numerical approaches to the correlated electron problem have achieved considerable success, yet are still constrained by several bottlenecks, including high order polynomial or exponential scaling in system size, long autocorrelation times, challenges in recognizing novel phases, and the Fermion sign problem. Methods in machine learning (ML), artificial intelligence, and data science promise to help address these limitations and open up a new frontier in strongly correlated quantum system simulations. In this paper, we review some of the progress in this area. We begin by examining these approaches in the context of classical models, where their underpinnings and application can be easily illustrated and benchmarked. We then discuss cases where ML methods have enabled scientific discovery. Finally, we will examine their applications in accelerating model solutions in state-of-the-art quantum many-body methods like quantum Monte Carlo and discuss potential future research directions.