A5043515839- Machine Learning in Materials Science
- X-ray Diffraction in Crystallography
- Catalytic Processes in Materials Science
- Electronic and Structural Properties of Oxides
- 2D Materials and Applications
- Advanced Battery Technologies Research
- Catalysis and Oxidation Reactions
- Advanced Battery Materials and Technologies
- CO2 Reduction Techniques and Catalysts
- Advanced Photocatalysis Techniques
- Fuel Cells and Related Materials
- Genomics and Chromatin Dynamics
- Gaussian Processes and Bayesian Inference
- Advanced X-ray and CT Imaging
- Advanced Materials Characterization Techniques
- RNA Research and Splicing
- Graphene research and applications
- Computational Drug Discovery Methods
- Machine Learning and Algorithms
- Radiomics and Machine Learning in Medical Imaging
- Advancements in Battery Materials
- Advanced Physical and Chemical Molecular Interactions
- Protein Structure and Dynamics
- Electron and X-Ray Spectroscopy Techniques
- Nuclear Structure and Function
Toyota Research Institute
2020-2025
Harvard University
2018-2024
Toyota Motor North America (United States)
2023
Harvard University Press
2022
Toyota Transportation Research Institute
2020
National Institute of Standards and Technology
2016
University of Rochester
2016
Abstract Machine learned force fields typically require manual construction of training sets consisting thousands first principles calculations, which can result in low efficiency and unpredictable errors when applied to structures not represented the set model. This severely limits practical application these models systems with dynamics governed by important rare events, such as chemical reactions diffusion. We present an adaptive Bayesian inference method for automating interpretable,...
We introduce configuration space as a natural representation for calculating the mechanical relaxation patterns of incommensurate two-dimensional (2D) bilayers, bypassing supercell approximations to encompass aperiodic patterns. The approach can be applied wide variety 2D materials through use continuum model in combination with generalized stacking fault energy interlayer interactions. present computational results small-angle twisted bilayer graphene and molybdenum disulfide (MoS$_2$),...
Abstract This work examines challenges associated with the accuracy of machine-learned force fields (MLFFs) for bulk solid and liquid phases d -block elements. In exhaustive detail, we contrast performance force, energy, stress predictions across transition metals two leading MLFF models: a kernel-based atomic cluster expansion method implemented using sparse Gaussian processes (FLARE), an equivariant message-passing neural network (NequIP). Early present higher relative errors are more...
Abstract X-ray absorption spectroscopy (XAS) produces a wealth of information about the local structure materials, but interpretation spectra often relies on easily accessible trends and prior assumptions structure. Recently, researchers have demonstrated that machine learning models can automate this process to predict coordinating environments absorbing atoms from their XAS spectra. However, are difficult interpret, making it challenging determine when they valid whether consistent with...
Abstract Rational catalyst design is crucial toward achieving more energy-efficient and sustainable catalytic processes. Understanding modeling reaction pathways kinetics require atomic level knowledge of the active sites. These structures often change dynamically during reactions are difficult to decipher. A prototypical example hydrogen-deuterium exchange catalyzed by dilute Pd-in-Au alloy nanoparticles. From a combination activity measurements, machine learning-enabled spectroscopic...
We use a data-driven approach to study the magnetic and thermodynamic properties of van der Waals (vdW) layered materials. investigate monolayers form [Formula: see text], based on known material using density functional theory (DFT) calculations machine learning methods determine their properties, such as order moment. also examine formation energies them proxy for chemical stability. show that tools, combined with DFT calculations, can provide computationally efficient means predict...
The modus operandi in materials research and development is combining existing data with an understanding of the underlying physics to create test new hypotheses via experiments or simulations. This process traditionally driven by subject expertise creativity individual researchers, who “close loop” updating their models light knowledge acquired from community. Since early 2000s, there has been notable progress automation each step scientific process. With recent advances using machine...
In this perspective, we highlight results of a research consortium devoted to advancing understanding oxygen reduction reaction (ORR) catalysis as means inform fuel cell science.
The oxygen vacancy formation energy (ΔEvf) governs defect concentrations alongside the entropy and is a useful metric to perform materials selection for variety of applications. However, density functional theory (DFT) calculations ΔEvf come at greater computational cost than typical bulk available in databases due involvement multiple vacancy-containing supercells. As result, repositories direct remain relatively scarce, development machine-learning models capable delivering accurate...
Quantum confinement endows two-dimensional (2D) layered materials with exceptional physics and novel properties compared to their bulk counterparts. Although certain two- few-layer configurations of graphene have been realized studied, a systematic investigation the arbitrarily assemblies is still lacking. We introduce theoretical concepts methods for processing information, as case study, apply them investigate electronic structure multi-layer graphene-based in high-throughput fashion....
Sequential learning for materials discovery is a paradigm where computational agent solicits new data to simultaneously update model in service of exploration (finding the largest number that meet some criteria) or exploitation with an ideal figure merit). In real-world campaigns, acquisition may be costly and optimal strategy involve using acquiring different levels fidelity, such as first-principles calculation supplement experiment. this work, we introduce agents which can operate on...
Exploratory synthesis has been the main generator of new inorganic materials for decades. AI-assisted discovery is possible, but human-AI collaboration should be refined according to their respective strengths.
Abstract Photoelectrocatalysts that use sunlight to power the CO 2 reduction reaction will be crucial for carbon-neutral and energy-efficient industrial processes. Scalable photoelectrocatalysts must satisfy a stringent set of criteria, such as stability under operating conditions, product selectivity, efficient light absorption. Two-dimensional materials can offer high specific surface area, tunability, potential heterostructuring, providing fresh landscape candidate catalysts. From...
Machine learning (ML) is gaining popularity as a tool for materials scientists to accelerate computation, automate data analysis, and predict properties. The representation of input material features critical the accuracy, interpretability, generalizability data-driven models scientific research. In this Perspective, we discuss few central challenges faced by ML practitioners in developing meaningful representations, including handling complexity real-world industry-relevant materials,...
Accurate predictions of the properties transition metal oxides using density functional theory (DFT) calculations are essential for computational design energy materials. In this work, we investigate anomalous reversal stability structural distortions (where distorted structures go from being energetically favorable to sharply unfavorable relative undistorted ones) induced by DFT+U on Mo d-orbitals in layered AMoO$_2$ (A = Li, Na, K) and rutile-like MoO$_2$. We highlight significant impact...
Data-driven interpretation of battery degradation visually summarizes the relationship between 16 state-of-health metrics and aging, facilitating users in simplifying large datasets identifying key regimes for further experimentation.
Molecular dynamics simulations are useful tools to screen solid polymer electrolytes with suitable properties applicable Li-ion batteries. However, due the vast design space of polymers, it is highly desirable accelerate screening by reducing computational time ion transport from simulations. In this study, we show that a judicious choice descriptors can predict equilibrium in LiTFSI–homopolymer systems within first 0.5 ns production run Specifically, find include information about behavior...
Abstract Forming a hetero-interface is materials-design strategy that can access an astronomically large phase space. However, the immense space necessitates high-throughput approach for optimal interface design. Here we introduce computational framework, InterMatch, efficiently predicting charge transfer, strain, and superlattice structure of by leveraging databases individual bulk materials. Specifically, algorithm reads in lattice vectors, density states, stiffness tensors each material...
Hundreds of genes interact with the yeast nuclear pore complex (NPC), localizing at periphery and clustering co-regulated genes. Dynamic tracking peripheral shows that they cycle on off NPC interaction slows their sub-diffusive movement. Furthermore, NPC-dependent inter-chromosomal leads to coordinated movement pairs loci separated by hundreds nanometers. We developed fractional Brownian motion simulations for chromosomal in nucleoplasm interacting NPCs. These predict rate nature random...