PINNs-hyperbolic-PDEs

Simple implementation of Physics Informed Neural Networks for some systems of hyperbolic PDEs using Pytorch

---

Currently implemented systems

• 1D Inviscid Burgers' equation

  

• 1D Shallow Water Equations with flat bottom

  • Entropy solution

  • Entropy violating solution

• 1D Acoustics with variable coefficients

  • Forward problem 2 homogeneous materials

  • Forward problem 4 homogeneous materials with few collocation points and just providing initial and boundary data

  • Forward problem 4 homogeneous materials with more collocation points and some data at the interior of the domain.

  • Inverse problem 2 homogeneous materials providing some data at the interior of the domain

---

Attributes

https://github.com/maziarraissi/PINNs

https://github.com/jayroxis/PINNs

https://github.com/clawpack

Please take a look at the file ATTRIBUTE.md for more details.

---

Requirements

My specific setup is provided in the file requirements.txt, but the following Python libraries are required

• NumPy
• Matplotlib
• PyTorch
• pyDOE
• scipy
• Jupyter Notebook
• CLAWPACK (If you want to set up a custom problem, data is provided for each system otherwise)