Hayden Hollenbeck

General surface properties to import: * model * reflectivity MPV * transmittance MPV * optical material 1 ID * optical material 2 ID GLISUR model: * polish * finish UNIFIED...

@stognini Here's a mapping of Geant4's models and finishes onto our layered surfaces. Celeritas options: Facet normal calculators: - Polished: facet normal is same as surface normal (no sampling) -...

Follow-up from our discussion today on possible interaction results: `InitBoundaryAction` set as post-step action when the track is geometry step limited and has been already moved to the boundary. Geometry...

I think it's best if we just use layer ID for the sense and have - pre-volume: layer = 0 - post-volume: layer = 1 and then it's basically just...

I believe yes, Geant4 does behave rationally for skin surfaces with painting. In the G4OpBoundaryProcess.cc most of that weird if statement chaining with polished/groundbackpainted is to handle whether the photon...

In classic physicist fashion, I think by describing them as coordinate frames we just need a single direction flag. ## Surface Record Frame The surface record frame is how the...

@sethrj I've been seeing some errors with finding `sincos` in the `GaussianRoughnessSampler` when trying to build some tests. Not sure what the regression was, maybe a missing include?

Sorry I was looking at the build-docker tests: ``` [429/1146] Building CUDA object src/celeritas/CMakeFiles/celeritas.dir/optical/surface/model/GaussianRoughnessModel.cu.o FAILED: src/celeritas/CMakeFiles/celeritas.dir/optical/surface/model/GaussianRoughnessModel.cu.o /usr/local/cuda/bin/nvcc -forward-unknown-to-host-compiler -Dceleritas_EXPORTS -I/__w/celeritas/celeritas/src -I/__w/celeritas/celeritas/build/include -isystem /usr/local/cuda/targets/x86_64-linux/include -isystem /opt/view/include -isystem /opt/software/linux-rocky9-x86_64_v3/gcc-11.5.0/openmpi-5.0.6-qzolp62fnhdcym2hkvnbvxujqzgeqtf2/include -Werror all-warnings -g...

Added some changes to simplify the applier code, use maps as inputs to models, and cache the track traversal direction. This should remove most of the ambiguities about whether vectors...

@sethrj @whokion you might find this interesting