How alexandredabyseesaram/neurom works
NeuROM is a refactoring-stage HiDeNN/FENNI finite-element neural-network library for interpretable parametric reduced-order modeling. Its documented workflow represents a field on a mesh, interpolates it using finite-element shape functions, and uses variational physics to train the model; constrained nodal values support Dirichlet conditions, while trainable geometry enables mesh adaptation. The graph includes the sampled mesh, field, mapping, quadrature-context and variational-term layers. Several central modules are named in the tree but were not sampled, so their internal wiring is left unspecified.
Main entry points: quadrature_context.py Quadrature context · mesh.py Mesh · term.py Variational terms · trainable_field.py Trainable field
Components 21
Fields and constraints
- FEM model
src/neurom/fem_model.py - Field
src/neurom/fields/field.py - Trainable field
src/neurom/fields/trainable_field.py - Constraints
src/neurom/constraints/constraint.py - Dirichlet BC
src/neurom/constraints/dirichlet.py
Mesh and geometry
- Mesh
src/neurom/meshes/mesh.py - Topology
src/neurom/meshes/topology.py - Isoparametric mapping
src/neurom/geometry/iso_parametric_mapping_1d.py - Shape functions
src/neurom/shape_functions/shape_function.py - Reference element
src/neurom/reference_elements/reference_element.py
Interpolation and quadrature
- Field layout
src/neurom/field_layout.py - Quadrature context
src/neurom/interpolation/quadrature_context.py - Quadrature rule
src/neurom/quadratures/quadrature_rule.py - Quadrature assembly
src/neurom/interpolation/quadrature_assembly.py
Variational physics
- Physics loss
src/neurom/physics_loss.py - Variational terms
src/neurom/physics/term.py - Elastic energy
src/neurom/physics/elastic_energy.py - Load potential
src/neurom/physics/load_potential.py - Field derivatives
src/neurom/differential.py - Integration
src/neurom/integrate.py
Other components
- Model user
Connections 14
- Model user → FEM model: configures
- Trainable field → Constraints: applies
- Dirichlet BC → Constraints: implements
- Trainable field → Topology: indexes with
- Mesh → Topology: uses
- Mesh → Field: positions from
- Isoparametric mapping → Shape functions: evaluates
- Quadrature context → Quadrature rule: uses
- Quadrature context → Mesh: reads
- Quadrature context → Isoparametric mapping: maps positions
- Quadrature rule → Reference element: references
- Variational terms → Field layout: reads
- Elastic energy → Variational terms: specializes
- Load potential → Variational terms: specializes