Picongpu: Compile error upon changing precision in collisions.param

[wmles]

concerning the newly merged PR #3416 - actually the changes that resolved issue #3461

in collisions.param the parameter float_COLL for the precision can be set. My paramset compiles with this value set to float_64 (the default), but the compilation breaks when changing that value to float_X or float_32.
This seems to be a minor issue, since using float_64 is beneficial anyways, but just wanted to mention.

The error is:
error: no suitable user-defined conversion from "pmacc::math::Vector<picongpu::precision32Bit::float_X, 3, pmacc::math::StandardAccessor, pmacc::math::StandardNavigator, pmacc::math::detail::Vector_components>" to "const picongpu::particles::collision::precision::float3_COLL" exists

my complete output from pic-build:
compile.log

[psychocoderHPC]

@pordyna I assigned you. I think it will be not so hard to fix.

[psychocoderHPC]

I will fix it. The problem is that the precision is hardcoded in the unitless file.

https://github.com/ComputationalRadiationPhysics/picongpu/blob/a9306d2eb2dfd7c93503dd48a507fda91b35c83c/include/picongpu/unitless/collision.unitless#L39

[pordyna]

Oh, missed that. Thank you! :smile:

[psychocoderHPC]

The bug should be solved with #3585, I was able to reproduce the issue with tests/CollisionsBeamRelaxation

[psychocoderHPC]

closed with #3585