How rishab-bansal/nvidia_physicsnemo_projects works
This repository contains two PhysicsNeMo examples that train physics-informed neural networks: a time-dependent 1D wave equation and a steady 2D lid-driven cavity flow. In each, a main script defines the physics and geometry, builds a network, and registers boundary and interior constraints before invoking the solver. The wave example also adds initial conditions and analytical-reference validation. Existing output trees show saved checkpoints and sampled constraint or validator data. The README is empty; relationships reflect the sampled scripts, not unsampled framework internals.
Main entry points: wave_1D_main.py Wave training · cavity_main.py Cavity training · wave_equation.py Wave equation · config.yml Wave config
Components 15
1D Wave
- Wave training
1D Wave Equation/wave_1D_main.py - Wave equation
1D Wave Equation/wave_equation.py - Wave config
1D Wave Equation/config.yml - Wave constraints
1D Wave Equation/wave_1D_main.py - Analytical validation
1D Wave Equation/wave_1D_main.py - Wave run artifacts
1D Wave Equation/outputs/wave_1D_main
Cavity Flow
- Cavity training
Cavity Flow/cavity_main.py - Flow equations
Cavity Flow/cavity_main.py - Cavity constraints
Cavity Flow/cavity_main.py - Cavity config
Cavity Flow/config.yml - Cavity run artifacts
Cavity Flow/outputs/cavity_main
PhysicsNeMo Runtime
- Network factory
- Domain and geometry
- Solver
Other components
- Engineer
Connections 17
- Engineer → Wave training: runs
- Engineer → Cavity training: runs
- Wave training → Wave config: loads config
- Wave training → Wave equation: constructs PDE
- Wave training → Network factory: instantiates network
- Wave training → Domain and geometry: builds domain
- Wave training → Wave constraints: registers constraints
- Wave training → Analytical validation: registers validator
- Wave training → Solver: creates and runs
- Solver → Wave run artifacts: writes artifacts
- Cavity training → Cavity config: loads config
- Cavity training → Flow equations: constructs equations
- Cavity training → Network factory: instantiates network
- Cavity training → Domain and geometry: builds domain
- Cavity training → Cavity constraints: registers constraints
- Cavity training → Solver: creates and runs
- Solver → Cavity run artifacts: writes artifacts