CFD PLATFORM
Open-source finite-volume CFD for fluids and continuum mechanics.
OPEN COMPUTATIONAL PHYSICS
Build, compare, and
understand PDE solvers.
A shared home for open-source numerical software, neural PDE frameworks, mesh-free methods, reproducible benchmarks and practical learning.
Classical solvers · Neural PDE frameworks · Mesh-free methods
PHYSICS IN VIEW01 / FLUID DYNAMICS
∇ · u = 0U∞
Different methods. Shared physics.Analytical potential flow · illustrative field
Speed / U∞
012
EXPLORE THE COMMONS
Models, solvers, and tools.
Explore established numerical software and the growing neural PDE ecosystem.
CLASSICAL & MULTIPHYSICS
FEM FRAMEWORK
Automated finite element computing through variational formulations.
FEM LIBRARY
High-performance finite elements from laptops to heterogeneous HPC.
COUPLING LIBRARY
Partitioned multiphysics coupling between existing solvers.
NEURAL PDE FRAMEWORKS
SCIENTIFIC ML
PINNs, DeepONets and multifidelity learning for differential equations.
PHYSICS-ML
Physics-ML models, training workflows and engineering examples.
OPERATOR LEARNING
PyTorch implementations of Fourier and other neural operators.
No matching resources. Try a method such as FEM or PINN, or clear your search.
9 resources shownSample catalogue · upstream projects remain independent · no affiliation implied
HARDWARE, WITH EVIDENCE
Performance means
more than speed.
Compare time and memory at the accuracy your problem requires. From a laptop to a cluster, see the complete computational cost.
- Shared problems and explicit error targets
- Training, adaptation and inference separated
- Reproducible software and hardware records
BENCHMARK EXPLORER
ILLUSTRATIVE DATATime to an accepted solution
Target relative L² error≤ 1%
Hardware classWall timePeak memory
0100 s
Example layout only. Values are synthetic—not measurements or hardware recommendations. Each row represents an illustrative solver configuration.
FROM PAPERS TO PRACTICE
Research that others can build on.
Implement the idea
A method becomes a reusable module, with its assumptions, dependencies and original paper.
Code + provenanceReproduce the result
Publish a runnable reference case and explain what matches—and what differs.
Reference case + evidenceMake it interoperable
Define interfaces and compatibility tests so others can compose it into new workflows.
Adapters + shared interfacesA foundation for industry-specific tools. Build blade-design, thermal-analysis or structural-engineering applications on shared computational capabilities.
LEARN BY DOING