The current implementation of sparse particle storage only works when the current state is stored as a vector of particles. For GPU implementations where the particles are stored as a dense matrix this isn't appropriate.
This can be overcome by adding a specialised getindex for these containers.
The current implementation of sparse particle storage only works when the current state is stored as a vector of particles. For GPU implementations where the particles are stored as a dense matrix this isn't appropriate.
This can be overcome by adding a specialised
getindex
for these containers.