Features

NetDEM is a versatile, neural network-enabled C++ library specifically designed for performing discrete element method (DEM) simulations. Its robust feature set includes:

• Sphere and Triangle Facet Contact Solver: A solver designed for accurate and efficient modeling of interactions between spherical particles and triangular surface facets.

• GJK (Gilbert-Johnson-Keerthi) Contact Solver for Convex Particles: This solver efficiently handles contact detection between convex particles, utilizing the GJK algorithm to provide rapid collision detection and response.

• SDF (Signed Distance Function) Contact Solver for Arbitrary Particles: The SDF contact solver supports both convex and concave particles, allowing for simulation of a broad range of particle geometries. This makes it especially useful for complex, irregularly shaped particles encountered in realistic scenarios.

• Hybrid OpenMP and MPI Parallel Computing: NetDEM uses hybrid parallel computing approaches to optimize performance. OpenMP is employed for multi-threading on shared-memory architectures, while MPI supports parallelism across distributed systems. This hybrid approach enables efficient handling of large-scale simulations.

• Integrated mlpack Machine Learning Environment: With mlpack integration, NetDEM users can leverage machine learning techniques directly within DEM simulations. This enables data-driven decision-making and predictive modeling to improve simulation efficiency and analysis.

NetDEM supports a wide range of particle shapes, including spheres, cylinders, poly-super-ellipsoids, poly-super-quadrics, spherical harmonics, triangle meshes, level sets, and more. This extensive shape support allows for flexible, highly customizable simulations applicable across a range of scientific and engineering applications.

License & Citation

NetDEM is distributed under the GPL license. See copyright and license for details.

See the about page for acknowledgements and citation information.