ParametricDMD

Parametric Dynamic Mode Decomposition for model discovery of dynamical systems.

This repository collects some code implementing Parametric Dynamic Mode Decomposition (Parametric DMD) algorithms as complementary material to the paper:

S. Riva, A. Missaglia, C. Introini, I. C. Bang, and A. Cammi, “A Comparison of Parametric Dynamic Mode Decomposition Algorithms for Thermal-Hydraulics Applications,” Mar. 2025. arXiv:2503.24205 [math]

The following test cases are provided:

1. Laminar flow over cylinder (Reynolds between 100 and 150) generated with dolfinx-v6 using OFELIA solvers
2. Flow over cylinder from CFDbench benchmark for Machine Learning
3. RELAP5 model of DYNASTY, a natural circulation loop deployed at Politecnico di Milano (see Riva et al. (2024) - NUTHOS14 Conference)

The data can be downloaded from Zenodo.

---

The code is written in Python and uses the following libraries:

• numpy
• scipy
• matplotlib
• pandas
• tqdm
• scikit-learn
• ezyrb
• pydmd https://github.com/PyDMD/PyDMD
• imageio.v2

---

Flow over cylinder - dolfinx-v6

Flow over cylinder - CFDbench

RELAP5 model of DYNASTY, natural circulation loop at Politecnico di Milano