Particle fields

Particle fields use the same cell-local storage as particle coordinates. Create them with init_cell_arrays so their slots and backend match particles:

pT, pρ, = init_cell_arrays(particles, Val(2))

init_cell_arrays(particles, Val(N)) returns a tuple of N scalar fields. The transfer routines accept such tuples directly, which is convenient for vector quantities:

pV = init_cell_arrays(particles, Val(2))
particle2grid!((Vx, Vy), pV, particles)

Pass companion fields through movement and injection routines so values stay attached to their particles:

move_particles!(particles, (pT, pρ))
inject_particles!(particles, (pT, pρ))

Only active particle slots contain valid values. Unused slots are represented by the occupancy/index mask and must not be included in reductions.

For per-cell data that is not one value per particle slot, such as one value per material phase in each cell, allocate the CellArray with cell_array:

n_phases = 3
phase_data = cell_array(backend, 0.0, (n_phases,), size(particles.index))

See CellArrays for indexing and halo exchange, and to_cpu for copying fields to host memory.