DClaw

A navigation hub for Physical AI and Scientific Machine Learning: systematically indexing five resource families — datasets & benchmarks, tools & frameworks, industrial cases, learning paths and papers & surveys — plus company research reports, a glossary, a weekly arXiv digest and a knowledge graph. Bilingual search and community submissions supported.

Browse Resources →

🗃️ Datasets & Benchmarks

Training & evaluation datasets, scientific ML benchmarks

🧰 Tools & Frameworks

Solvers, training frameworks, post-processing & visualization tools

🎯 Applied Cases

Physics AI deployment cases in any domain - CAE, energy, weather, biomedicine, materials, geophysics and beyond

🎓 Learning Paths

Resources from beginner to industrial deployment

📄 Papers & Surveys

Must-read benchmark papers and technical surveys (no news or industry reports)

Featured

PDEBench

PDE Benchmark / Multi-Equation

Large-scale PDE benchmark covering heat, advection-diffusion and fluid dynamics - supports fair comparison of data-driven and physics-informed methods.

MIT License BenchmarkPDECFD

The Well

Multiphysics / PDE - 16 Domains

Unified benchmark spanning 16 physical domains, dubbed the 'ImageNet moment of physical ML', with PyTorch API, pretrained baselines and HF streaming.

Per subset (mostly CC-BY / code MIT) MultiphysicsBenchmarkPyTorch

DeepXDE

PINN Starter Framework

The most beginner-friendly PINN framework, with excellent Chinese docs and TensorFlow/PyTorch backends - the top choice to start with physics-informed networks.

GNU Lesser General Public License v2.1 PINNBeginnerChinese Docs

OpenFOAM

General-Purpose CFD Solver

The de facto open-source CFD solver covering incompressible/compressible flow, heat transfer, multiphase and combustion - the workhorse for generating physics-AI training data.

GPL v3 CFDOpen SourceSolver

PhysicsNeMo

Solver / Training Framework

NVIDIA's physics-ML training framework (formerly Modulus) with PINN, neural operators and physics-constrained training - commercial-friendly, industry-first choice.

Apache-2.0 PINNNeural OperatorNVIDIA

GraphCast: Data-Driven Weather Forecasting

Weather / Large-Scale Physical Prediction

A graph neural network for medium-range weather prediction that outperforms conventional numerical weather prediction on many targets while running in minutes instead of hours.

arXiv preprint (copyright held by authors) WeatherGraph Neural NetworkForecasting

MACE-MP-0: A Universal Machine-Learned Interatomic Potential

Molecular Dynamics / ML Interatomic Potential

A general-purpose machine-learned interatomic potential built on an equivariant graph neural network, trained on 150k public inorganic crystals, that runs stable molecular dynamics out of the box for solids, liquids, gases and reactions, and can be fine-tuned to ab initio accuracy with only a handful of configurations.

MIT (code) / arXiv preprint Molecular DynamicsML PotentialEquivariant GNN

Physics-Based Deep Learning (Online Book)

Textbook / Systematic Course

A free online textbook from TU Munich covering physics + deep learning: differentiable physics, PINNs, neural operators and turbulence modeling, with runnable code.

CC BY-NC-SA 4.0 TextbookDifferentiable PhysicsPINN

DeepONet (Original Paper)

Foundational Paper

Presents a branch-trunk architecture realizing a deep learning version of the universal operator approximation theorem - the first systematic architecture for learning nonlinear operators and still a mainstream baseline.

arXiv preprint (copyright held by authors) DeepONetNeural OperatorFoundational

Recent Updates

ClimateLearn: Benchmarking Machine Learning for Weather and Climate

Climate ML Benchmarking Library

ClimateLearn (UCLA Machine Intelligence Group, 2023) is a Python library that provides standardized access to climate datasets, baseline models, and metrics for large-scale benchmarking of statistical downscaling and temporal forecasting methods. Code is MIT licensed.

New MIT ClimateMachine LearningBenchmark

GROMACS

Biomolecular Molecular Dynamics

A high-performance molecular dynamics package focused on biomolecules (proteins, lipids, nucleic acids), renowned for extreme parallelism and vectorization — a mainstay of structural biology and drug design.

New GNU Lesser General Public License v2.1 Molecular DynamicsBiomolecularProtein

JAX

Differentiable Numerical Computing Framework

Google's open-source NumPy-style numerical computing library built on XLA with autodiff and functional transformations (grad/jit/pmap/vmap) — the substrate behind physics-AI differentiable solvers and neural operators like JAX-CFD and JAX-Fluids.

New Apache License 2.0 Automatic DifferentiationXLANumerical Computing

LAMMPS

Molecular Dynamics & Many-Body Simulation

A Sandia-led open-source molecular dynamics and particle simulator supporting multi-scale models from atomic to coarse-grained and continuum, with a large library of force fields — the workhorse for materials, chemistry and biomolecular simulation.

New GNU General Public License v2.0 Molecular DynamicsMany-Body SimulationForce Field

NVIDIA Warp

Differentiable GPU Simulation Framework

NVIDIA's open-source Python framework that JIT-compiles plain Python functions into CPU/CUDA kernels with automatic differentiable adjoints, letting gradients flow through entire physics simulations.

New Apache License 2.0 Differentiable SimulationGPUJIT Compilation

PyTorch Geometric (PyG)

Graph Neural Network Framework

A PyTorch-based library for graph neural networks offering rich GNN operators, popular datasets and samplers — the de facto standard for graph representation learning on molecules, crystals and physical systems.

New MIT License Graph Neural NetworkGNNPyTorch

Taichi Lang

Physics Simulation DSL

A Python-embedded DSL for high-performance computing that turns Python into parallel, auto-differentiable kernels, with built-in MPM/SPH/FEM frontends for physical simulation.

New Apache License 2.0 Physical SimulationDifferentiable ProgrammingDSL

AlphaFold2 — Deep Learning Breakthrough in Protein Structure Prediction

Protein Structure Prediction

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.

New Apache-2.0 (code) / CC-BY-4.0 (parameters) Protein StructureDeep LearningStructural Biology

FourCastNet — Global Weather Model with Adaptive Fourier Neural Operators

Weather Forecasting

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.

New BSD 3-Clause License (code) Weather ForecastingFourier Neural OperatorDeep Learning

GenCast: Diffusion-based Ensemble Weather Forecasting

Probabilistic Ensemble Weather Forecasting

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).

New Apache-2.0（代码）/ CC-BY-NC-SA-4.0（模型权重，禁止商用） Weather ForecastingDeep LearningProbabilistic Forecasting

NeuralGCM: Hybrid ML + Physics Atmospheric General Circulation Model

Hybrid Atmospheric General Circulation Model

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.

New Apache-2.0（代码）/ CC-BY-SA-4.0（模型权重） Weather ForecastingClimate SimulationHybrid Model

OpenFold: Trainable PyTorch Reproduction of AlphaFold2