This is a continuation of my masters project. It will provide proof of concept for estimation of the distribution function of a non-equilibrium hydrodynamic fluid cell. The function finding will take place via learning the momentum distribution from the transport picture of particles, where we can also compute hydrodynamic state variables. We will arrive at a mapping from non-equilibrium state variables, such as shear and bulk stress, to a possible distribution function. The connection of state variables to distribution function can be learned via a neural network, which is the main goal of this project.

I will continue to work on this in my free time, connecting the pure parameter scan with learning via neural networks. So this project is under construction and will undergo heavy changes in the next weeks.

• Maxwell Jüttner Data Sampler ✓
• Plot particles in momentum space ✓
• Update Explanation for physics ✓
• Document steps until now better ✓
• Plot momentum in space ✓
• Calculate the energy-momentum tensor ✓
• Plot distribution function
• Distort momentum space to incorporate either Bulk or shear stress transformation
• Plot distribution function again
• Calculate non-equilibrium quantities

• Find suitable NN to handle data
  • translational invariance of energy-momentum tensor --> CNN?

. . .

• Clean README for release state