Clearly differentiate between participant_dt and dt

[BenjaminRodenberg]

I think we should be more clear here, because otherwise the double meaning of dt can lead to bugs that are difficult to find. During the preCICE workshop 2023 I found a bug in the code of a user where the statement dt = beginTimeStep() was left out, because dt was fixed to a constant before the timeloop started dt = subcycling_dt. This is totally understandable from my perspective, if you do not use adaptivity. Why should I redefine dt in every time step, if it stays the same?

The goal of the user was to use subcycling. However, this led to the situation that dt was getting smaller than subcycling_dt when getting close to the end of the first window: min(precice_dt, subcycling_dt) returns precice_dt < subcycling_dt and we overwrite dt leading to dt < subcycling_dt for the remainder of the simulation time.

I think this is a situation that we can avoid easily and at the same time we can show clearly that there are basically three different dts:

1. dt which is the dt that we actually use for the time step
2. precice_dt which is the maximum allowed dt defined by preCICE
3. participant_dt (or solver_dt?) which is the dt desired by the solver (fixed or computed), but might conflict with precice_dt

This is still a draft, because we still have the following todos:

1. Do we want this change?
2. Apply it consistently to figures, solverdummies etc.

[uekerman]

Yes, why not. I am only afraid that this will be "high effort – low gain".

[BenjaminRodenberg]

@uekerman Any preference participant_dt or solver_dt? I think both is fine, but would use solver_dt, because it's shorter and we also write double dt; // solver timestep size in the step-by-step guide.

[uekerman]

Any preference participant_dt or solver_dt?

No, not really. We explain that both are valid options here: https://precice.org/fundamentals-terminology.html#solver-and-participant

[BenjaminRodenberg]

@uekerman I hope I did not oversee any necessary changes.

One thing I just stumbled over: precice_dt or dt_precice? Same for solver_dt. The text in https://precice.org/couple-your-code-timestep-sizes.html sometimes mentions dt_precice. Thinking about the mathematical notation $\Delta t_\text{preCICE}$, dt_precice would probably make more sense. Same for $\delta t_\mathcal{D}$ in https://precice.org/couple-your-code-waveform.html.

Note: Our tutorials currently use precice_dt or my_dt (e.g. here)

Pragmatic suggestion: Let's keep precice_dt, because I think this is really only about consistency (and means a lot of work) and it does not really influence readability.

[fsimonis]

Just for completeness: Once we rename the SolverInterface to Participant precice/precice#1458, then using participant_dt for the solver would be highly confusing.

[uekerman]

Looks good