Wiley Interdisciplinary Reviews: Computational Molecular Science最新文献

Deep learning has deeply changed the paradigms of many research fields. At the heart of chemical and physical sciences is the accurate ab initio calculation of many-body wavefunctions, which has become one of the most notable examples to demonstrate the power of deep learning in science. In particular, the introduction of deep learning into quantum Monte Carlo (QMC) has significantly advanced the frontier of ab initio calculation, offering a universal tool to solve the electronic structure of materials and molecules. Deep learning QMC architectures were initially designed and tested on small molecules, focusing on comparisons with other state-of-the-art ab initio methods. Methodological developments, including extensions to real solids and periodic models, have been rapidly progressing, and reported applications are fast expanding. This review covers the theoretical foundation of deep learning QMC for solids, the neural network wavefunction ansatz, and various other methodological developments. Applications on computing energy, electron density, electric polarization, force, and stress of real solids are also reviewed. The methods have also been extended to other periodic systems and finite temperature calculations. The review highlights the potential and existing challenges of deep learning QMC in materials chemistry and condensed matter physics.

With the increasing global demand for energy transition and environmental sustainability, catalysts play a vital role in mitigating global climate change, as they facilitate over 90% of chemical and material conversions. It is important to investigate the complex structures and properties of catalysts for enhanced performance, for which artificial intelligence (AI) methods, especially graph neural networks (GNNs) could be useful. In this article, we explore the cutting-edge applications and future potential of GNNs in intelligent catalyst design. The fundamental theories of GNNs and their practical applications in catalytic material simulation and inverse design are first reviewed. We analyze the critical roles of GNNs in accelerating material screening, performance prediction, reaction pathway analysis, and mechanism modeling. By leveraging graph convolution techniques to accurately represent molecular structures, integrating symmetry constraints to ensure physical consistency, and applying generative models to efficiently explore the design space, these approaches work synergistically to enhance the efficiency and accuracy of catalyst design. Furthermore, we highlight high-quality databases crucial for catalysis research and explore the innovative application of GNNs in thermocatalysis, electrocatalysis, photocatalysis, and biocatalysis. In the end, we highlight key directions for advancing GNNs in catalysis: dynamic frameworks for real-time conditions, hierarchical models linking atomic details to catalyst features, multi-task networks for performance prediction, and interpretability mechanisms to reveal critical reaction pathways. We believe these advancements will significantly broaden the role of GNNs in catalysis science, paving the way for more efficient, accurate, and sustainable catalyst design methodologies.

Protein–protein interactions play a crucial role in human biological processes, and deciphering their structural information and interaction patterns is essential for drug development. The high costs of experimental structure determination have brought computational protein–protein docking methods into the spotlight. Traditional docking algorithms, which hinge on a sampling-scoring framework, heavily rely on extensive sampling of candidate poses and customized scoring functions based on the geometric and chemical compatibility between proteins. However, these methods face challenges related to sampling efficiency and stability. The advent of deep learning (DL) has ushered in data-driven docking methods that demonstrate significant advantages, particularly boosting the efficiency of protein–protein docking. We systematically review the historical development of protein–protein docking from traditional approaches to DL techniques and provide insights into emerging technologies in this field. Moreover, we summarize the commonly used datasets and evaluation metrics in protein–protein docking. We expect that this review can offer valuable guidance for the development of more efficient protein–protein docking algorithms.

Recent advances in single-atom alloy (SAA) catalysts provide a unique platform for understanding spillover, due to the well-defined nature of the active site for dissociative chemisorption. In particular, the use of spilled adsorbates following molecular dissociation on the host metal surface facilitates the generation of high-value chemicals in subsequent catalytic reactions. Nevertheless, the factors that control the spillover process remain to be fully elucidated. This perspective discusses recent theoretical advances in the spillover dynamics on SAAs, with a particular focus on the dissociation and spillover processes of H₂ and CH₄. It provides valuable insights into how various factors, such as energy transfer, nuclear quantum effects, gas-adsorbate interactions, and adsorbate size, impact the diffusion behavior of hydrogen and methyl species on SAA surfaces. The article concludes with a discussion of future prospects. This perspective underscores the significance of spillover dynamics in heterogeneous catalysis, with important implications for improving catalytic performance.

We describe our contribution as industrial stakeholders to the existing open-source GPU4PySCF project (https://github.com/pyscf/gpu4pyscf), a GPU-accelerated Python quantum chemistry package. We have integrated GPU acceleration into other PySCF functionalities including Density Functional Theory (DFT), geometry optimization, frequency analysis, solvent models, and the density fitting technique. Through these contributions, GPU4PySCF v1.0 can now be regarded as a fully functional and industrially relevant platform, which we demonstrate in this work through a range of tests. When performing DFT calculations with the density fitting scheme on modern GPU platforms, GPU4PySCF delivers a 30 times speedup over a 32-core CPU node, resulting in approximately 90% cost savings for most DFT tasks. The performance advantages and productivity improvements have been found in multiple industrial applications, such as generating potential energy surfaces, analyzing molecular properties, calculating solvation free energy, identifying chemical reactions in lithium-ion batteries, and accelerating neural-network methods. With the improved design that makes it easy to integrate with the Python and PySCF ecosystem, GPU4PySCF is a natural choice that we can now recommend for many industrial quantum chemistry applications.

Transition-metal compounds (TMCs) with open-shell d-electrons are characterized by a complex interplay of lattice, charge, orbital, and spin degrees of freedom, giving rise to various fascinating applications. Often exhibiting exotic properties, these compounds are commonly classified as correlated systems due to strong inter-electronic interactions called Hubbard U. This inherent complexity presents significant challenges to Kohn-Sham density functional theory (KS-DFT), the most widely used electronic structure method in condensed matter physics and materials science. While KS-DFT is, in principle, exact for the ground-state total energy, its exchange-correlation energy must be approximated in practice. The mean-field nature of KS implementations, combined with the limitations of current exchange-correlation density functional approximations, has led to the perception that DFT is inadequate for correlated systems, particularly TMCs. Consequently, a common workaround involves augmenting DFT with an on-site Hubbard-like U correction. In recent years, the strongly constrained and appropriately normed (SCAN) density functional, along with its refined variant r²SCAN, has achieved remarkable progress in accurately describing the structural, energetic, electronic, magnetic, and vibrational properties of TMCs, challenging the traditional perception of DFT's limitations. This review explores the design principles of SCAN and r²SCAN, highlights their key advancements in studying TMCs, explains the mechanisms driving these improvements, and addresses the remaining challenges in this evolving field.

Molecular dynamics (MD) simulations have become a powerful tool for studying double-layer systems, offering atomistic insights into their equilibrium properties and dynamic behaviors. These simulations have significantly advanced the understanding of key electrochemical mechanisms and the design of electrochemical devices. However, challenges remain in aligning simulations with the complexities of realistic applications. In this perspectiv, we highlight critical areas for enhancing the realism of MD simulations, including refining methods for representing electrode polarization, improving electrode and electrolyte models to incorporate structural and compositional complexities, and simulating charging and discharging processes under realistic conditions while considering associated thermal behaviors. We also stress the importance of scaling simulation results to experimental dimensions through multiscale modeling and dimensionless analysis. Overcoming these challenges will allow MD simulations to advance our understanding of electrical double-layer behaviors and drive innovations in the development of future electrochemical technologies.

The catalytic CO₂ hydrogenation to produce valuable fuels and chemicals holds immense importance in addressing energy scarcity and environmental degradation. Given that the real catalytic reaction system is complex and dynamic, the structure of catalysts might experience dynamic evolution under real reaction conditions. It implies that the real active sites might only generated during the reaction process. The induction factor of dynamic evolution of active sites could be reactants, intermediates, products, and other local chemical environments. Utilizing in-situ/operando characterization techniques allows for the real-time observation of the dynamic evolution process, further combining multiscale theoretical simulations can effectively reveal the refined structure of real active sites and catalytic mechanisms. Herein, we summarized the latest advancements in understanding the dynamic active sites and catalytic mechanisms during the real reaction process for the CO₂ hydrogenation to C₁ products (CH₃OH, CO, and CH₄). The dynamic evolutions of the catalyst in morphology, size, valence state, and interface between active component and support were discussed, respectively. Future research could benefit from more in-situ characterization and theoretical simulation to explore the microstructure and reaction mechanism, aiming to produce high conversion and selectivity catalysts for CO₂ hydrogenation reactions.

Molecular quantum chemistry has seen enormous progress in the last few decades thanks to more advanced and sophisticated numerical techniques and computing power. Following the recent interest in extending these capabilities to condensed-phase problems, we summarize basic knowledge of condensed-phase quantum chemistry for readers with experience in molecular quantum chemistry. We highlight recent efforts in this direction, including solving the electron repulsion integrals bottleneck, implementing hybrid density functional theory and wavefunction methods, and simulating lattice dynamics for periodic systems within atom-centered basis sets. Many computational techniques presented here are inspired by the extensive method developments rooted in quantum chemistry. In this Focus Article, we selectively focus on the computational techniques rooted in molecular quantum chemistry, emphasize some challenges, and point out open questions. We hope our perspectives will encourage researchers to pursue this exciting and promising research avenue.

In recent decades, coupled cluster theory has proven valuable in accurately describing correlation in many-body systems, particularly in time-independent computations of molecular electronic structure and vibrations. This review describes recent advancements in using coupled cluster parameterizations for time-dependent wave functions for the efficient computation of the quantum dynamics associated with the motion of nuclei. It covers time-dependent vibrational coupled cluster (TDVCC) and time-dependent modal vibrational coupled cluster (TDMVCC), which employ static and adaptive basis sets, respectively. We discuss the theoretical foundation, including many-mode second quantization, bivariational principles, and various parameterizations of time-dependent bases. Additionally, we highlight key features that make TDMVCC promising for future quantum dynamical simulations. These features include fast configuration-space convergence, the use of a compact adaptive basis set, and the possibility of efficient implementations with a computational cost that scales only polynomially with system size.