Chengyan StudioComputational Research Services

机器学习与AI

Machine Learning and AI

Connect first-principles, molecular dynamics and continuum modeling with data-driven methods, covering machine-learning potentials, property prediction and screening, generative design and optimization, and multiscale surrogates with limited-scope digital twins.

Machine-Learning Potentials and Large-Scale Simulation

ML Potential Construction and Active Learning

ML Potential Construction and Active Learning

ML interatomic potential, active learning, uncertainty sampling, DFT labels, DeePMD, MACE, NequIP, Allegro, energy and force errors

Universal Potential Assessment and Fine-Tuning

Universal Potential Assessment and Fine-Tuning

universal MLIP, MACE-MP, CHGNet, DPA, zero-shot evaluation, fine-tuning, force error, property validation

Machine-Learning Potential Molecular Dynamics

Machine-Learning Potential Molecular Dynamics

MLMD, long-timescale simulation, large systems, diffusion coefficient, interface reconstruction, reaction sampling, comparison with AIMD

Property Prediction and High-Throughput Screening

Structure–Property Models and High-Throughput Screening

Structure–Property Models and High-Throughput Screening

graph neural network, descriptor model, formation energy, adsorption energy, band gap, high-throughput screening, cross-validation, DFT verification

Interpretable Models and Symbolic-Regression Descriptors

Interpretable Models and Symbolic-Regression Descriptors

symbolic regression, SISSO, sparse descriptor, feature importance, structure–activity relation, interpretable machine learning

Data Mining and Machine Learning Modeling

Data Mining and Machine Learning Modeling

multi-source data integration, data cleaning, denoising, missing-value handling, outlier detection, feature engineering, dimensionality reduction, regression / classification, cross-validation, SHAP interpretability, literature data extraction, spectral / cycling time-series data

Generative Design and Intelligent Optimization

Generative Models and Inverse Design

Generative Models and Inverse Design

generative model, inverse design, crystal generation, molecular-skeleton design, property constraints, DFT stability check

Bayesian Optimization and Computational Experiment Design

Bayesian Optimization and Computational Experiment Design

Bayesian optimization, Gaussian process, experiment design, composition-space search, computable objective, iterative recommendation

Multiscale Surrogates and Digital Twins

Surrogate Models for FEM and Phase-Field

Surrogate Models for FEM and Phase-Field

surrogate model, phase-field acceleration, FEM surrogate, reduced-order model, parameter sweep, error assessment

Data Assimilation and Model Parameter Identification

Data Assimilation and Model Parameter Identification

data assimilation, Bayesian calibration, parameter inversion, phase-field, finite element, fusion of observations, uncertainty

Battery Electrochemical–Thermal Model and State Estimation

Battery Electrochemical–Thermal Model and State Estimation

electrochemical–thermal coupling, parameter identification, state estimation, operating-condition prediction, cycling data, limited-scope digital twin

Physics-Informed Neural Networks

Physics-Informed Neural Networks

PINN, physics constraints, sparse data, continuum model, conservation constraints, extrapolation risk

Accelerated Electronic Structure and Hamiltonian Learning

Accelerated Electronic Structure and Hamiltonian Learning

learned DFT Hamiltonian, DeepH, hybrid-functional acceleration, band structure, density of states, comparison with DFT