speed comparison of Particle Swarm with user-defined objective function (hopefully parallel)?

[yezhengli-Mr9]

1. speed comparison of Particle Swarm with user-defined objective function? With other particle swarm implementation (preferably c++)

2. even particles within same swarm, not parallel implementation? Although saw thread-safe discussion like Passing arguments in a user defined problem #559 (comment)

pagmo2/src/algorithms/pso.cpp

Lines 229 to 237 in ac7b6d2

	// For each particle in the swarm
	for (decltype(swarm_size) p = 0u; p < swarm_size; ++p) {
	// identify the current particle's best neighbour
	// . not needed if m_neighb_type == 1 (gbest): best_neighb directly tracked in this function
	// . not needed if m_variant == 6 (FIPS): all neighbours are considered, no need to identify the best
	// one
	if (m_neighb_type != 1u && m_variant != 6u)
	best_neighb = particle__get_best_neighbor(p, neighb, lbX, lbfit);

Also checked sample codes for basis cases (without user defined objective nor UD classes involved)
https://esa.github.io/pagmo2/docs/cpp/algorithms/pso.html
https://github.com/esa/pagmo2/blob/master/tests/pso_gen.cpp
https://github.com/esa/pagmo2/blob/master/tests/pso.cpp

pagmo2/tests/pso.cpp

Lines 151 to 156 in ac7b6d2

	// Make one evolution
	problem prob{rosenbrock{25u}};
	population pop{prob, 5u, 23u};
	algorithm algo{pso{500u, 0.79, 2., 2., 0.1, 5u, 2u, 4u, false, 23u}};
	algo.set_verbosity(23u);
	pop = algo.evolve(pop);

Thanks