Uncertainty-driven dynamics for active learning of interatomic potentials

Nature Computational Science, Год журнала: 2023, Номер 3(3), С. 230 - 239

Опубликована: Март 6, 2023

Machine learning (ML) models, if trained to data sets of high-fidelity quantum simulations, produce accurate and efficient interatomic potentials. Active (AL) is a powerful tool iteratively generate diverse sets. In this approach, the ML model provides an uncertainty estimate along with its prediction for each new atomic configuration. If passes certain threshold, then configuration included in set. Here we develop strategy more rapidly discover configurations that meaningfully augment training The uncertainty-driven dynamics active (UDD-AL), modifies potential energy surface used molecular simulations favor regions space which there large uncertainty. performance UDD-AL demonstrated two AL tasks: sampling conformational glycine promotion proton transfer acetylacetone. method shown efficiently explore chemically relevant space, may be inaccessible using regular dynamical at target temperature conditions.

Язык: Английский

---

Nanocrystal Assemblies: Current Advances and Open Problems

ACS Nano, Год журнала: 2024, Номер 18(23), С. 14791 - 14840

Опубликована: Май 30, 2024

We explore the potential of nanocrystals (a term used equivalently to nanoparticles) as building blocks for nanomaterials, and current advances open challenges fundamental science developments applications. Nanocrystal assemblies are inherently multiscale, generation revolutionary material properties requires a precise understanding relationship between structure function, former being determined by classical effects latter often quantum effects. With an emphasis on theory computation, we discuss that hamper assembly strategies what extent nanocrystal represent thermodynamic equilibrium or kinetically trapped metastable states. also examine dynamic optimization protocols. Finally, promising functions examples their realization with assemblies.

Язык: Английский

---

Machine learning interatomic potential: Bridge the gap between small-scale models and realistic device-scale simulations

iScience, Год журнала: 2024, Номер 27(5), С. 109673 - 109673

Опубликована: Апрель 4, 2024

Machine learning interatomic potential (MLIP) overcomes the challenges of high computational costs in density-functional theory and relatively low accuracy classical large-scale molecular dynamics, facilitating more efficient precise simulations materials research design. In this review, current state four essential stages MLIP is discussed, including data generation methods, material structure descriptors, six unique machine algorithms, available software. Furthermore, applications various fields are investigated, notably phase-change memory materials, searching, properties predicting, pre-trained universal models. Eventually, future perspectives, consisting standard datasets, transferability, generalization, trade-off between complexity MLIPs, reported.

Язык: Английский

---

Stability Oracle: a structure-based graph-transformer framework for identifying stabilizing mutations

Nature Communications, Год журнала: 2024, Номер 15(1)

Опубликована: Июль 23, 2024

Engineering stabilized proteins is a fundamental challenge in the development of industrial and pharmaceutical biotechnologies. We present Stability Oracle: structure-based graph-transformer framework that achieves SOTA performance on accurately identifying thermodynamically stabilizing mutations. Our introduces several innovations to overcome well-known challenges data scarcity bias, generalization, computation time, such as: Thermodynamic Permutations for augmentation, structural amino acid embeddings model mutation with single structure, protein structure-specific attention-bias mechanism makes transformers viable alternative graph neural networks. provide training/test splits mitigate leakage ensure proper evaluation. Furthermore, examine our engineering contributions, we fine-tune ESM2 representations (Prostata-IFML) achieve sequence-based models. Notably, Oracle outperforms Prostata-IFML even though it was pretrained 2000X less has 548X parameters. establishes path fine-tuning virtually any phenotype, necessary task accelerating protein-based

Язык: Английский

---

Decoding Electrochemical Processes of Lithium‐Ion Batteries by Classical Molecular Dynamics Simulations

Advanced Energy Materials, Год журнала: 2024, Номер 14(22)

Опубликована: Март 19, 2024

Abstract Lithium‐ion batteries (LIBs) have played an essential role in the energy storage industry and dominated power sources for consumer electronics electric vehicles. Understanding electrochemistry of LIBs at molecular scale is significant improving their performance, stability, lifetime, safety. Classical dynamics (MD) simulations could directly capture atomic motions thus provide dynamic insights into electrochemical processes ion transport during charging discharging that are usually challenging to observe experimentally, which momentous developing with superb performance. This review discusses developments MD approaches using non‐reactive force fields, reactive machine learning potential modeling chemical reactions reactants electrodes, electrolytes, electrode‐electrolyte interfaces. It also comprehensively how interactions, structures, transport, reaction affect electrode capacity, interfacial properties. Finally, remaining challenges envisioned future routes commented on high‐fidelity, effective simulation methods decode invisible interactions LIBs.

Язык: Английский

---

Data Generation for Machine Learning Interatomic Potentials and Beyond

Chemical Reviews, Год журнала: 2024, Номер 124(24), С. 13681 - 13714

Опубликована: Ноя. 21, 2024

The field of data-driven chemistry is undergoing an evolution, driven by innovations in machine learning models for predicting molecular properties and behavior. Recent strides ML-based interatomic potentials have paved the way accurate modeling diverse chemical structural at atomic level. key determinant defining MLIP reliability remains quality training data. A paramount challenge lies constructing sets that capture specific domains vast space. This Review navigates intricate landscape essential components integrity data ensure extensibility transferability resulting models. We delve into details active learning, discussing its various facets implementations. outline different types uncertainty quantification applied to atomistic acquisition correlations between estimated true error. role samplers generating informative structures highlighted. Furthermore, we discuss via modified surrogate potential energy surfaces as innovative approach diversify also provides a list publicly available cover

Язык: Английский

---

Computational Methods for Modeling Lipid-Mediated Active Pharmaceutical Ingredient Delivery

Molecular Pharmaceutics, Год журнала: 2025, Номер unknown

Опубликована: Янв. 29, 2025

Lipid-mediated delivery of active pharmaceutical ingredients (API) opened new possibilities in advanced therapies. By encapsulating an API into a lipid nanocarrier (LNC), one can safely deliver APIs not soluble water, those with otherwise strong adverse effects, or very fragile ones such as nucleic acids. However, for the rational design LNCs, detailed understanding composition-structure-function relationships is missing. This review presents currently available computational methods LNC investigation, screening, and design. The state-of-the-art physics-based approaches are described, focus on molecular dynamics simulations all-atom coarse-grained resolution. Their strengths weaknesses discussed, highlighting aspects necessary obtaining reliable results simulations. Furthermore, machine learning, i.e., data-based approach to lipid-mediated introduced. data produced by experimental theoretical provide valuable insights. Processing these help optimize LNCs better performance. In final section this Review, computer reviewed, specifically addressing compatibility

Язык: Английский

---

AIMNet2: A Neural Network Potential to Meet your Neutral, Charged, Organic, and Elemental-Organic Needs

Опубликована: Окт. 12, 2023

Machine learned interatomic potentials (MLIPs) are reshaping computational chemistry practices because of their ability to drastically exceed the accuracy-length/time scale tradeoff. Despite this attraction, benefits such efficiency only impactful when an MLIP uniquely enables insight into a target system or is broadly transferable outside training dataset, where models achieving latter seldom reported. In work, we present 2nd generation our atoms-in-molecules neural network potential (AIMNet2), which applicable species composed up 14 chemical elements in both neutral and charged states, making it valuable model for modeling majority non-metallic compounds. Using exhaustive dataset 20 million hybrid quantum calculations, AIMNet2 combines ML-parameterized short-range physics-based long-range terms attain generalizability that reaches from simple organics diverse molecules with “exotic” element-organic bonding. We show outperforms semi-empirical GFN-xTB on par reference density functional theory interaction energy contributions, conformer search tasks, torsion rotation profiles, molecular-to-macromolecular geometry optimization. Overall, demonstrated coverage significant step toward providing access MLIPs avoid crucial limitation curating additional data retraining each new application.

Язык: Английский

---

Applications of machine learning in supercritical fluids research

The Journal of Supercritical Fluids, Год журнала: 2023, Номер 202, С. 106051 - 106051

Опубликована: Июль 25, 2023

Язык: Английский

---

Machine learning electronic structure methods based on the one-electron reduced density matrix

Nature Communications, Год журнала: 2023, Номер 14(1)

Опубликована: Окт. 7, 2023

The theorems of density functional theory (DFT) establish bijective maps between the local external potential a many-body system and its electron density, wavefunction and, therefore, one-particle reduced matrix. Building on this foundation, we show that machine learning models based one-electron matrix can be used to generate surrogate electronic structure methods. We surrogates hybrid DFT, Hartree-Fock full configuration interaction theories for systems ranging from small molecules such as water more complex compounds like benzene propanol. use central quantity learned. From predicted matrices, either standard quantum chemistry or second machine-learning model compute molecular observables, energies, atomic forces. essentially anything method can, band gaps Kohn-Sham orbitals energy-conserving ab-initio dynamics simulations infrared spectra, which account anharmonicity thermal effects, without need employ computationally expensive algorithms self-consistent field theory. are packaged in an efficient easy Python code, QMLearn, accessible popular platforms.

Язык: Английский

---