`compute_smoothed_velocity!` is dispatched incorrectly

[efaulhaber]

TrixiParticles.jl/src/schemes/boundary/wall_boundary/dummy_particles.jl

Lines 662 to 672 in 9755a09

	@propagate_inbounds function compute_smoothed_velocity!(cache, viscosity::ViscosityAdami,
	neighbor_system, v_neighbor_system,
	kernel_weight, particle, neighbor)
	v_b = current_velocity(v_neighbor_system, neighbor_system, neighbor)
	for dim in eachindex(v_b)
	cache.wall_velocity[dim, particle] += kernel_weight * v_b[dim]
	end
	return cache
	end

Am I missing something, or is this function redundant? It stores something in the wall_velocity, but inside boundary_pressure_extrapolation!. After that, we call compute_adami_density!, which computes the inverse state equation and calls compute_wall_velocity!, which overwrites wall_velocity.

@LasNikas @svchb can you please double-check before I remove this function?

[LasNikas]

IIRC (without looking into the code):
compute_smoothed_velocity just interpolates the fluid velocity to the boundary particle locations.
compute_wall_velocity! then calculates the mirrored boundary velocity (therby enforcing the no-slip condition through symmetric viscous coupling).

[efaulhaber]

Yes, but the smoothed velocity calculated in compute_smoothed_velocity! is not used in compute_wall_velocity!.

[efaulhaber]

Yes, but the smoothed velocity calculated in compute_smoothed_velocity! is not used in compute_wall_velocity!.

Which is correct btw. $v_a$ in the paper is just the wall velocity, nothing smoothed or interpolated.

[LasNikas]

TrixiParticles.jl/src/schemes/boundary/wall_boundary/dummy_particles.jl

Lines 689 to 692 in 9755a09

	new_velocity = @inbounds 2 * v_boundary[dim] -
	wall_velocity[dim, particle] / volume[particle]
	@inbounds wall_velocity[dim, particle] = new_velocity
	end

It is used in L690?

[svchb]

I think its not overwritten if no viscosity is set.

@inline function compute_wall_velocity!(viscosity::Nothing, system, system_coords, particle)
    return viscosity
end

[LasNikas]

I think its not overwritten if no viscosity is set.

It is then not calculated at all

[efaulhaber]

It is used in L690?

Ah, you're right. That is correct then.

But it is not correct that compute_smoothed_velocity! is dispatched on ViscosityAdami. This should happen for every viscosity model.

[LasNikas]

But it is not correct that compute_smoothed_velocity! is dispatched on ViscosityAdami. This should happen for every viscosity model.

Yes, also the naming is very confusing.

[efaulhaber]

the naming is very confusing.

I think it's good. What would you change? Maybe rename compute_smoothed_velocity! to interpolate_fluid_velocity!? And add more comments.