A compositional, energy-based software framework for modeling mechanical, electromagnetic and thermodynamic systems
EPHS.jl lets you represent physical systems as composable energy-flow diagrams grounded in the port-Hamiltonian approach. Crucially, the graphical syntax and the port-Hamiltonian semantics compose in compatible ways, enabling seamless hierarchical (de)composition of systems. Additional structural properties guarantee not just passivity, but thermodynamic consistency (1st & 2nd law). The primary goal is to make complex energy-based models easier to build, understand, reuse, and refactor without sacrificing mathematical rigor.
- Hierarchical model structure to manage complexity
- Simple reuse of models at all levels of abstraction
- Thermodynamic consistency follows recursively from consistency of subsystems
📘 Online documentation:
https://MarkusLohmayer.github.io/EPHS.jl
- Extend the underlying computer algebra system to multiple theories (Lie groups and algebras, exterior calculus)
- Add capabilities for optimization of symbolic expressions and theorem proofing using (slotted) e-graphs
- Provide tooling for data-driven parameter identification
- Implement various numerical integrators as natural transformations
- Demonstrate that the port-Hamiltonian modeling language provides a well-suited target language for generative AI in physical modeling