A5060737716- Theoretical and Computational Physics
- Machine Learning in Materials Science
- Protein Structure and Dynamics
- Stochastic processes and statistical mechanics
- Material Dynamics and Properties
- Computational Drug Discovery Methods
- Advanced Chemical Physics Studies
- Computational Physics and Python Applications
- Neural Networks and Applications
- Physics of Superconductivity and Magnetism
- Complex Network Analysis Techniques
- Statistical Mechanics and Entropy
- Parallel Computing and Optimization Techniques
- Mass Spectrometry Techniques and Applications
- Advanced Neural Network Applications
- Cloud Computing and Resource Management
- Distributed and Parallel Computing Systems
- Chemical Synthesis and Analysis
- Multiferroics and related materials
- Advanced Data Storage Technologies
- Quantum and electron transport phenomena
- Polymer Surface Interaction Studies
- Quantum many-body systems
- Markov Chains and Monte Carlo Methods
- Ferroelectric and Piezoelectric Materials
Los Alamos National Laboratory
2019-2025
Chengdu University of Information Technology
2024
Liaoning Technical University
2024
Hangzhou Dianzi University
2024
Sichuan University
2024
Jilin University
2024
Changsha University of Science and Technology
2023
Oak Ridge National Laboratory
2013-2021
China University of Mining and Technology
2017
National Transportation Research Center
2016
QMCPACK is an open source quantum Monte Carlo package for ab initio electronic structure calculations. It supports calculations of metallic and insulating solids, molecules, atoms, some model Hamiltonians. Implemented real space algorithms include variational, diffusion, reptation Carlo. uses Slater–Jastrow type trial wavefunctions in conjunction with a sophisticated optimizer capable optimizing tens thousands parameters. The orbital auxiliary-field method also implemented, enabling cross...
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...
We introduce a parallel Wang-Landau method based on the replica-exchange framework for Monte Carlo simulations. To demonstrate its advantages and general applicability simulations of complex systems, we apply it to different spin models including glasses, Ising model Potts model, lattice protein adsorption, self-assembly process in amphiphilic solutions. Without loss accuracy, gives significant speed-up potentially scales up petaflop machines.
Abstract Complex behavior poses challenges in extracting models from experiment. An example is spin liquid formation frustrated magnets like Dy 2 Ti O 7 . Understanding has been hindered by issues including disorder, glass formation, and interpretation of scattering data. Here, we use an automated capability to extract model Hamiltonians data, identify different magnetic regimes. This involves training autoencoder learn a compressed representation three-dimensional diffuse scattering, over...
Machine learning (ML) is quickly becoming a premier tool for modeling chemical processes and materials. ML-based force fields, trained on large data sets of high-quality electron structure calculations, are particularly attractive due their unique combination computational efficiency physical accuracy. This Perspective summarizes some recent advances in the development neural network-based interatomic potentials. Designing training crucial to overall model One strategy active learning, which...
Insulators play a pivotal role in power transmission lines, and the timely detection of defects insulators is crucial to prevent potentially catastrophic consequences terms human lives property. This paper proposes an insulator defect algorithm, named Insulator Lack-You Only Look Once (IL-YOLO), addressing limitations observed existing research concerning complex background multi-target challenges detection. The IL-YOLO algorithm focuses on detecting within intricate lines. To enhance its...
Methodologies for training machine learning potentials (MLPs) with quantum-mechanical simulation data have recently seen tremendous progress. Experimental a very different character than simulated data, and most MLP procedures cannot be easily adapted to incorporate both types of into the process. We investigate procedure based on iterative Boltzmann inversion that produces pair potential correction an existing using equilibrium radial distribution function data. By applying these...
We investigate a generic, parallel replica-exchange framework for Monte Carlo simulations based on the Wang-Landau method. To demonstrate its advantages and general applicability massively of complex systems, we apply it to lattice spin models, self-assembly process in amphiphilic solutions, adsorption molecules surfaces. While current interest, latter phenomena are challenging study computationally because multiple structural transitions occurring over broad temperature range. show how...
Abstract Using the van der Waals density functional with C09 exchange (vdW-DF-C09), which has been applied to describing a wide range of dispersion-bound systems, we explore physical properties prototypical AB O 3 bulk ferroelectric oxides. Surprisingly, vdW-DF-C09 provides superior description experimental values for lattice constants, polarization and moduli, exhibiting similar accuracy modified Perdew-Burke-Erzenhoff was designed specifically solids (PBEsol). The relative performance is...
Accurate large-scale first principles calculations based on density functional theory (DFT) in metallic systems are prohibitively expensive due to the asymptotic cubic scaling computational complexity with number of electrons. Using algorithmic advances employing finite-element discretization for DFT (DFT-FE) conjunction efficient methodologies and mixed precision strategies, we delay onset this by significantly reducing prefactor while increasing arithmetic intensity lowering data movement...
Sunny is a Julia package designed to serve the needs of quantum magnetism community. It supports specification very broad class spin models and diverse suite numerical solvers. These include powerful methods for simulating dynamics both in out equilibrium. Uniquely, it features generalization classical semiclassical approaches SU(N) coherent states, which useful studying systems exhibiting strong spin-orbit coupling or local entanglement effects. also offers well-developed framework...
Machine learning inter-atomic potentials (MLIPs) are revolutionizing the field of molecular dynamics (MD) simulations. Recent MLIPs have tended towards more complex architectures trained on larger datasets. The resulting increase in computational and memory costs may prohibit application these to perform large-scale MD Here, we present a teacher-student training framework which latent knowledge from teacher (atomic energies) is used augment students' training. We show that light-weight...
The growing interest in machine learning (ML) tools within chemistry and material science stems from their novelty ability to predict properties almost as accurately underlying electronic structure calculations or experiments. Transition path sampling (TPS) offers a practical way explore transition routes between metastable minima such conformers isomers on the multidimensional potential energy surface. However, TPS has historically suffered computational cost vs. accuracy trade-off...
The growing interest in machine learning (ML) tools within chemistry and material science stems from their novelty ability to predict properties almost as accurately underlying electronic structure calculations...
The thermodynamic behavior and structural properties of hydrophobic-polar (HP) lattice proteins interacting with attractive surfaces are studied by means Wang-Landau sampling. Three benchmark HP sequences (48mer, 67mer, 103mer) considered different types surfaces, each which attract either all monomers, only hydrophobic (H) or polar (P) respectively. diversity folding in dependence surface strength is discussed. Analyzing the combined patterns various observables, such as, e.g., derivatives...
Catalyzed by enormous success in the industrial sector, many research programs have been exploring data-driven, machine learning approaches. Performance can be poor when model is extrapolated to new regions of chemical space, e.g., bonding types, many-body interactions. Another important limitation spatial locality assumption architecture, and this cannot overcome with larger or more diverse datasets. The outlined challenges are primarily associated lack electronic structure information...
UAV aerial images of insulators have problems such as complex backgrounds, small targets, and obscured resulting in low detection accuracy insulator targets. In order to solve the above problems, a HRD-YOLOX(Hybrid Attention Mechanisms, Regularization, Depth Separable Convolution for YOLOX) algorithm suitable recognizing detecting defects is proposed based on YOLOX network. First, hybrid attention module HAM-CSP(Communication Sequential Processes Hybrid Mechanisms) incorporated into feature...
High-performance computing (HPC) increasingly relies on heterogeneous architectures to achieve higher performance.In Oak Ridge Leadership Facility (OLCF), this trend continues as its latest supercomputer, Summit, entered production in early 2019.The combination of IBM POWER 9 CPU and Nvidia V100 GPU, along with a fast NVLink2 interconnect other technologies, pushes system performance new height breakes the exascale barrier by certain measures.Due Summit's powerful GPUs much GPU-CPU ratio,...
The growing interest in machine learning (ML) tools within chemistry and material science stems from their novelty ability to predict properties almost as accurately underlying electronic structure calculations or experiments. Transition path sampling (TPS) offers a practical way explore transition routes between metastable minima such conformers isomers on the multidimensional potential energy surface. However, TPS has historically suffered computational cost vs. accuracy trade-off...
The authors provide a method to calculate systematically the effect of single-site mutations on thermodynamic and structural properties large hydrophobic-polar lattice protein models. Their provides way such as mean energy, heat capacity, ground-state population proteins larger than what is currently feasible with atomistic