DeepMind's 2021 end-to-end deep learning model that lifted protein 3D structure prediction to near-experimental accuracy at CASP14, transforming structural biology. Code Apache-2.0; parameters CC-BY-4.0.
NVIDIA et al.'s 2022 global data-driven high-resolution weather model at 0.25-degree resolution that generates a week-long forecast in seconds, matching ECMWF IFS accuracy on large-scale variables and enabling thousand-member ensemble forecasting. Code BSD-3-Clause.
GenCast is a generative ML weather model from Google DeepMind (2023) that uses a diffusion model on the spherical grid to produce probabilistic ensemble forecasts, surpassing the operational physics-based ensemble (ECMWF ENS) on most targets while running orders of magnitude faster. Code is Apache-2.0; model weights are CC-BY-NC-SA-4.0 (non-commercial).
NeuralGCM (Google DeepMind, 2024; Nature) is a hybrid ML + physics atmospheric model. A neural network parameterizes sub-grid physics while a differentiable dynamical core preserves numerical stability, combining physical consistency with data-driven accuracy. Code is Apache-2.0; weights are CC-BY-SA-4.0.
NewApache-2.0（代码）/ CC-BY-SA-4.0（模型权重）Weather ForecastingClimate SimulationHybrid Model
OpenFold (Columbia AQLab, 2022) is a faithful, trainable, memory-efficient and GPU-friendly PyTorch reproduction of DeepMind's AlphaFold2. Source code is Apache-2.0; DeepMind's pretrained parameters are CC-BY-4.0 (relaxed from CC BY-NC 4.0 as of January 2022).
Huawei Cloud's 2023 Nature model for 3D high-resolution global weather forecasting that consistently outperforms the ECMWF IFS baseline and was adopted by ECMWF for operational testing. Model weights CC-BY-NC-SA-4.0 (non-commercial).
NewCC-BY-NC-SA-4.0 (model weights; commercial use forbidden)Weather ForecastingDeep LearningMeteorology
RoseTTAFold (UW Baker lab, 2021; Science) predicts protein structures and interactions with a three-track network (1D sequence / 2D distance / 3D coordinates), performing strongly at CASP14. Code is MIT; trained weights and data are under the Rosetta-DL Software license for non-commercial use only.
NIST-led benchmark suite for additive manufacturing - standard test artifacts and gold-standard measurements (X-ray CT, synchrotron) validate melt-pool, residual-stress, and microstructure models for metal 3D printing.
NewPublic data (NIST public domain)Additive ManufacturingMetal 3D PrintingBenchmark
DOE/NREL building energy simulation engine solving hourly heat balances; neural surrogates replace repeated runs to open the feasible region for portfolio-scale design and operation optimization.
Caltech-led open-source chemical kinetics suite - homogeneous/heterogeneous reacting-flow solvers for combustion and electrochemistry; with ML potentials and surrogates it forms infrastructure for AI chemistry.
Industrial-scale case of ML potentials replacing DFT for electrocatalysis screening - OC20-trained models like GemNet accelerate adsorption-energy calculations by orders of magnitude.
The de-facto open-source CFD standard - finite-volume framework covering heat transfer, combustion, multiphase and electromagnetics; the usual data supplier for physics-AI training.
PEER-led open-source structural simulation framework - nonlinear fiber beams and 2-D solid elements for seismic analysis; combined with ML it powers rapid regional damage assessment workflows.
Representative case of neural surrogates for reservoir simulation - PhysicsNeMo examples predict pressure and saturation fields, serving history matching, production optimization and carbon-storage monitoring.
Open-source physics-based battery modelling library - multi-scale electrochemical models from single-particle to full cell, with parameter estimation and fast simulation for digital twins and SOH prediction.
NREL-led open-source wave energy converter simulator - Simulink/Simscape multibody modeling coupled with BEM/potential-flow solvers; standard tool for device design and control verification.
NVIDIA Earth-2 enables weather-sensitive industries like energy to make timely decisions by integrating observational data with AI-powered atmospheric simulation.
See LICENSE fileClimate SimulationWeather PredictionAtmospheric Physics
