Relying more on default parameters for the acceleration

[MakisH]

Originally posted by @uekerman in #640 (comment):

If we touch this, I would suggest to directly switch to the default everywhere, e.g.

<acceleration:IQN-ILS>
  <data name="Displacement" mesh="Solid-Mesh" />
  <data name="Force" mesh="Solid-Mesh" />
</acceleration:IQN-ILS>

This would mean, for the perpendicular-flap case:

    <acceleration:IQN-ILS>
      <data name="Displacement" mesh="Solid-Mesh" />
      <data name="Force" mesh="Solid-Mesh" />
-      <preconditioner type="residual-sum" update-on-threshold="false" />
-      <filter type="QR2" limit="1e-2" />
-      <initial-relaxation value="0.5" />
-      <max-used-iterations value="100" />
-      <time-windows-reused value="15" />
    </acceleration:IQN-ILS>

We could revisit all tutorials and remove all non-default values, or update to newer options (e.g., use the QR3 filter in more tutorials).

This would, however, require re-validation of each tutorial that is updated (and then updating the reference results).

Compare to the defaults in https://precice.org/configuration-xml-reference.html

[PranjalManhgaye]

hii , @MakisH i would like to work on this , ig this would be great to start , i checked the current "IQN-ILS" blocks and noticed that several tutorials use values that differ from the current defaults, for example "perpendicular-flap" uses "initial-relaxation=0.5" , "filter=QR2" , and "time-windows-reused=15" , while the current defaults are different

Before opening a PR, i want to clarify the intended scope:

1. Should we remove only redundant values that match current defaults?
2. Or should we intentionally switch selected tutorials to current defaults, accepting that this may require re-validation and updated reference results , it may affect numerical behaviour too ?

To keep it reviewable , i will start first with " perpendicular-flap " as first pr , since it is also mentioned in the issue as the example and then expand it , in this way we will be able to validate/reference each tutorials once the strategy is clear .

strategy would be to group similar tutorials together and expand gradually after the first PR.

[MakisH]

Thanks for the offer, @PranjalManhgaye.

Option 1 would be rather easy. I am not sure if we have such cases at the moment, but if we have, it would be nice and easy to clean them up. This, I would encourage you to try.

Option 2 will require significant effort in testing/validation and updating images: For example, the perpendicular-flap tutorial has several plots that need to be redone. I would not start that without a good first exploration, and I would start from tutorials with fewer solver options.

[PranjalManhgaye]

thanks @MakisH. i will start with option 1 and first audit the "IQN-ILS" blocks for values that exactly match the current defaults, without touching tuned/non-default values. i will open a small PR after checking the cases .