#include "lattice.h"

#include "LLBM/collide.h"
#include "LLBM/initialize.h"
#include "LLBM/bounce_back.h"
#include "LLBM/bounce_back_moving_wall.h"
#include "LLBM/equilibrium_density_wall.h"
#include "LLBM/equilibrium_velocity_wall.h"

#include <chrono>
#include <iostream>

#include "pattern/all.h"

using T = SWEEPLB_PRECISION;
using PATTERN = pattern::SWEEPLB_PATTERN<T>;

void simulate(Cuboid cuboid, std::size_t nStep) {
	const int nThread = omp_get_max_threads();

	Lattice<PATTERN> lattice(cuboid);

	LatticeMask<T> bulk_mask(cuboid.volume());
	LatticeMask<T> wall_mask(cuboid.volume());
	LatticeMask<T> inflow_mask(cuboid.volume());
	LatticeMask<T> outflow_mask(cuboid.volume());

	cuboid.traverse([&](int iX, int iY, int iZ, std::size_t iCell) {
		if (   iY == 0 || iY == cuboid[1]-1
		    || iZ == 0 || iZ == cuboid[2]-1) {
			wall_mask.set(iCell, true);
		} else if (iX == 0) {
			inflow_mask.set(iCell, true);
		} else if (iX == cuboid[0]-1) {
			outflow_mask.set(iCell, true);
		} else {
			bulk_mask.set(iCell, true);
		}
	});

	bulk_mask.serialize();
	wall_mask.serialize();
	inflow_mask.serialize();
	outflow_mask.serialize();

	T tau = 0.56;

	T u_inflow  = 0.05;
	T d_outflow = 1.;

	for (std::size_t iStep = 0; iStep < 100; ++iStep) {
		lattice.apply(Operator(BgkCollideO(), bulk_mask, tau),
		              Operator(BounceBackO(), wall_mask),
		              Operator(EquilibriumVelocityWallO(), inflow_mask, u_inflow, WallNormal<1,0,0>()),
		              Operator(EquilibriumDensityWallO(), outflow_mask, d_outflow, WallNormal<-1,0,0>()));
		lattice.stream();
	}

	auto start = std::chrono::steady_clock::now();

	for (std::size_t iStep = 0; iStep < nStep; ++iStep) {
		lattice.apply(Operator(BgkCollideO(), bulk_mask, tau),
		              Operator(BounceBackO(), wall_mask),
		              Operator(EquilibriumVelocityWallO(), inflow_mask, u_inflow, WallNormal<1,0,0>()),
		              Operator(EquilibriumDensityWallO(), outflow_mask, d_outflow, WallNormal<-1,0,0>()));
		lattice.stream();
	}

	auto duration = std::chrono::duration_cast<std::chrono::duration<double>>(
	  std::chrono::steady_clock::now() - start);

	std::cout << cuboid[0] << ", " << cuboid[1] << ", " << cuboid[2]
	          << ", " << nStep
	          << ", " << nThread
	          << ", " << (nStep * lattice.volume()) / (1e6 * duration.count())
	          << std::endl;

	lattice.write_momenta(bulk_mask, "result.vtk");
}

int main(int argc, char* argv[]) {
	const std::size_t nX    = atoi(argv[1]);
	const std::size_t nY    = atoi(argv[2]);
	const std::size_t nZ    = atoi(argv[3]);
	const std::size_t steps = atoi(argv[4]);
	simulate({ nX, nY, nZ }, steps);
}