Fusion Extension implementation - inter-species

[FilipO28555]

This extension is based on papers:
Wu et al
PIC code Chicago
Cannoni
Cross section parametrization:
Bosh, Hale

There is now documentation.

Code flow:

                                                               ┌ IntraCollision.hpp (not yet tested/included - will be in the next PR)
simulation.hpp -> Fusion.x.cpp -> Collider.hpp -> WithPeer.hpp ┤
                                                               └ InterCollision.hpp

Struct InterCollision{} is a combination of algorithm from Collision Extension and Creation Kernel.

                   uses: (particles/fusion/detail) 
InterCollision.hpp --------------------------------> FusionFunctor.hpp -> FusionAlgorithm.hpp

FusionAlgorithm.hpp decides if fusion happens and chooses momenta of new particles.

The setup that works and tests this extension is at: Fusion_Setups/DT-nHe_hal8999

Some preliminary results:

To do in consecutive PRs:

• Add ci test
• Add intra-species fusion
• More cross-section interpolators

[pordyna]

Right now this reuses a lot of collision code simply by copying files and also somewhat modifying them. We should unify this, for example, by generalizing the collision kernels. I'm not sure if this should happen before this is merged. @psychocoderHPC what do you think?

[psychocoderHPC]

Right now this reuses a lot of collision code simply by copying files and also somewhat modifying them. We should unify this, for example, by generalizing the collision kernels. I'm not sure if this should happen before this is merged. @psychocoderHPC what do you think?

we should keep the code first seperate, after merging we have a baseline to validate if we break something. Than we should check what we can unify.
This way requires that we not stop after after it is merged.
Before next week I am most likely not possible to look into this pr.

[BrianMarre]

@FilipO28555 the CI is still complaining that some source code files are marked as executable, for example,

• include/picongpu/particles/fusion/detail/ListEntry.hpp
• include/picongpu/particles/fusion/WithPeer.hpp

only binaries and scripts should be marked as executable, please follow the directions in the CI-job output on how to fix this.

[PrometheusPi]

@FilipO28555 This is not yet a full review, but it would be great if you remove unused /comment-out code so that the number of added files gets less scary (right now it's 3.5k lines)

[PrometheusPi]

Please remove uncommented /unused code

[ikbuibui]

One option if you want to keep this code is to add it as an example and activate it. That way we get better CI testing as well.

[FilipO28555]

OK, Now the fusion.param is a bare minimum implementation - I moved the examples to documentation

[ikbuibui]

I see you have really moved all the example code into the documentation.
We avoid doing this because it is easy to overlook updating the docs when something in your implementation changes, and we end up with code in our docs which doesn't compile.
The way we deal with this is to put the param file into an example/test set up under share/picongpu/examples or share/picongpu/tests and then include this code into the doxygen. This way the code is checked for compilation and if something in the code changes, you don't have to separately updated the docs.
You can look at this to see how to include code into doxygen.

[ikbuibui]

In general, we avoid merge commits
See here and here

[ikbuibui]

Also I'm not sure how/if possible it is anymore but it would have been useful to have a commit at the state where you simply copied code from the collision pipeline and made it compile, and then added the fusion changes on top with separate commits.

[BrianMarre]

@psychocoderHPC any progress on reviewing this PR?

[psychocoderHPC]

Please review your own changes a remove hard coded numbers e.g. 1.0 by the correct precision of the value e.g. 1.0_COLL to avoid silent implicit casts.

[psychocoderHPC]

please add a doc string and unit, additionally the type is an integral type but you assigned an double value.

[BrianMarre]

@psychocoderHPC a few CI jobs failed with due to script errors, please restart them

[PrometheusPi]

CI job restarted

[PrometheusPi]

I triggered the CI again, we have some connectivity issues:

fatal: unable to access 'https://github.com/ComputationalRadiationPhysics/isaac.git/': Could not resolve host: github.com