You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Graph neural network surrogates for crash simulation: autoregressive deformation prediction with MeshGraphNets and MGN-T, from raw FE data to evaluation, in a lean PyTorch framework. Evaluated on deforming_plate and a bumper beam crash.
Casting solidification GNN pipeline: STEP to healed CAD, Gmsh mesh, predicted solidification time, hot spots and shrinkage screening. The model is on Hugging Face (eugenmik/castsolid-gnn), Apache-2.0.
Trustworthy surrogate models for structural finite-element analysis: gated CalculiX data, shape-conditioned scalar and MeshGraphNet models, FastAPI service and React app.