Improve comments

[SCiarella]

Suggestions by @luisaforozco :

• CNODE maybe explicitly mention what C stands for coupled.
• Like that you keep a single way of defining the initial values.
• Not clear the use of ; in some cases, it doesn't seem consistent.
• In example2_part2 when plotting multi_GS.gif put labels of the 3 rows or explain that they are 3 different examples (i.e. with different parameters).
• In example2_part2 the coarser grid is only for v, not for u?
• Loos MulDtO not clear can we add a link to resources explaining it?
• Homogenize notation to what Benjamin and CWI time uses?
• Make it clear/distinction between a priori and a posteriori loss/train/inference.
• I would like to have an outline of what is going to be done in the example. For instance in example 2:
  1. Definition of the problem: discretization, initial conditions, parameters.
  2. Definition of elements of the equations: RHS, closure.
  3. Burn-out to initialize closure model.
  4. Generation of results. And for example2_part2
  5. Definition of the problem: discretization, initial conditions, parameters.
  6. Definition of elements of the equations: RHS, closure (NO NN yet).
  7. Burn-out to initialize closure model x2.
  8. Define a coarser grid.
  9. Get the solution for the coarser grid and compare with the solutions of the fine ones.

[luisaforozco]

We need to also:

• [x] Check the import statements so that the dependency tree is clear for everyone.

[APJansen]

Also:

• make it more clear that the training data comes from a NeuralODE because it's the analogue of DNS

[SCiarella]

also:

• [ ] more clear names to the different examples