/* Copyright 2017-2021 PaGMO development team

This file is part of the PaGMO library.

The PaGMO library is free software; you can redistribute it and/or modify
it under the terms of either:

  * the GNU Lesser General Public License as published by the Free
    Software Foundation; either version 3 of the License, or (at your
    option) any later version.

or

  * the GNU General Public License as published by the Free Software
    Foundation; either version 3 of the License, or (at your option) any
    later version.

or both in parallel, as here.

The PaGMO library is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
for more details.

You should have received copies of the GNU General Public License and the
GNU Lesser General Public License along with the PaGMO library.  If not,
see https://www.gnu.org/licenses/. */

#define BOOST_TEST_MODULE pso_gen_test
#define BOOST_TEST_DYN_LINK
#include <boost/test/unit_test.hpp>

#include <iostream>
#include <limits> //  std::numeric_limits<double>::infinity();
#include <numeric>
#include <string>

#include <boost/lexical_cast.hpp>
#include <boost/test/tools/floating_point_comparison.hpp>

#include <pagmo/algorithm.hpp>
#include <pagmo/algorithms/pso_gen.hpp>
#include <pagmo/population.hpp>
#include <pagmo/problems/hock_schittkowski_71.hpp>
#include <pagmo/problems/rosenbrock.hpp>
#include <pagmo/problems/zdt.hpp>
#include <pagmo/rng.hpp>

using namespace pagmo;

struct my_sto_prob {

    my_sto_prob(unsigned dim = 2u, unsigned sample_size = 10u, unsigned seed = pagmo::random_device::next())
        : m_dim(dim), m_sample_size(sample_size), m_e(seed), m_seed(seed)
    {
    }

    /// Sets the seed
    /**
     * @param seed the random number generator seed
     */
    void set_seed(unsigned seed)
    {
        m_seed = seed;
    }

    vector_double fitness(const vector_double &x) const
    {
        // We seed the random engine
        m_e.seed(m_seed);
        // We construct a uniform distribution from 0 to 1.
        auto drng = std::normal_distribution<double>(0., 1.0);
        auto x_copy = x;

        std::transform(x.begin(), x.end(), x_copy.begin(), [](const double &el) { return std::pow(el, 2.); });
        double retval = 0;

        for (auto i = 0u; i < m_sample_size; ++i) {
            auto noise = drng(m_e);
            retval += noise;
        }
        double spam = std::pow(std::accumulate(x_copy.begin(), x_copy.end(), 0.0), 1. / 5.);
        retval = (retval / m_sample_size) + spam;
        return {retval};
    }

    /// Box-bounds
    /**
     *
     * It returns the box-bounds for this UDP.
     *
     * @return the lower and upper bounds for each of the decision vector components
     */
    std::pair<vector_double, vector_double> get_bounds() const
    {
        vector_double lb(m_dim, -100.);
        vector_double ub(m_dim, 100.);
        return {lb, ub};
    }

    // problem dimension
    unsigned m_dim = 2u;
    // sample size
    unsigned m_sample_size = 10u;
    // Random engine
    mutable detail::random_engine_type m_e;
    // Seed
    unsigned m_seed;
};

BOOST_AUTO_TEST_CASE(construction)
{
    BOOST_CHECK_NO_THROW(pso_gen{});
    pso_gen user_algo{100, 0.79, 2., 2., 0.1, 5u, 2u, 4u, false, 23u};
    BOOST_CHECK(user_algo.get_verbosity() == 0u);
    BOOST_CHECK(user_algo.get_seed() == 23u);
    BOOST_CHECK((user_algo.get_log() == pso_gen::log_type{}));

    BOOST_CHECK_NO_THROW((pso_gen{100, 0.79, 2., 2., 0.1, 5u, 2u, 4u, false, 23u}));

    BOOST_CHECK_THROW((pso_gen{100, -0.79, 2., 2., 0.1, 5u, 2u, 4u, false, 23u}), std::invalid_argument);
    BOOST_CHECK_THROW((pso_gen{100, 2.3, 2., 2., 0.1, 5u, 2u, 4u, false, 23u}), std::invalid_argument);

    BOOST_CHECK_THROW((pso_gen{100, 0.79, -1., 2., 0.1, 5u, 2u, 4u, false, 23u}), std::invalid_argument);
    BOOST_CHECK_THROW((pso_gen{100, 0.79, 2., -1., 0.1, 5u, 2u, 4u, false, 23u}), std::invalid_argument);
    BOOST_CHECK_THROW((pso_gen{100, 0.79, 5., 2., 0.1, 5u, 2u, 4u, false, 23u}), std::invalid_argument);
    BOOST_CHECK_THROW((pso_gen{100, 0.79, 2., 5., 0.1, 5u, 2u, 4u, false, 23u}), std::invalid_argument);

    BOOST_CHECK_THROW((pso_gen{100, 0.79, 2., 2., -2.3, 5u, 2u, 4u, false, 23u}), std::invalid_argument);
    BOOST_CHECK_THROW((pso_gen{100, 0.79, 2., 2., 1.1, 5u, 2u, 4u, false, 23u}), std::invalid_argument);

    BOOST_CHECK_THROW((pso_gen{100, 0.79, 2., 2., 0.1, 8u, 2u, 4u, false, 23u}), std::invalid_argument);
    BOOST_CHECK_THROW((pso_gen{100, 0.79, 2., 2., 0.1, 0u, 2u, 4u, false, 23u}), std::invalid_argument);

    BOOST_CHECK_THROW((pso_gen{100, 0.79, 2., 2., 0.1, 5u, 6u, 4u, false, 23u}), std::invalid_argument);
    BOOST_CHECK_THROW((pso_gen{100, 0.79, 2., 2., 0.1, 5u, 0u, 4u, false, 23u}), std::invalid_argument);

    BOOST_CHECK_THROW((pso_gen{100, 0.79, 2., 2., 0.1, 5u, 2u, 0u, false, 23u}), std::invalid_argument);
}

BOOST_AUTO_TEST_CASE(evolve_test)
{
    // We then check that the evolve throws if called on unsuitable problems
    BOOST_CHECK_THROW(pso_gen{10u}.evolve(population{problem{rosenbrock{}}}), std::invalid_argument);
    BOOST_CHECK_THROW(pso_gen{10u}.evolve(population{problem{zdt{}}, 15u}), std::invalid_argument);
    BOOST_CHECK_THROW(pso_gen{10u}.evolve(population{problem{hock_schittkowski_71{}}, 15u}), std::invalid_argument);
    // And a clean exit for 0 generations
    population pop{rosenbrock{2u}, 20u};
    BOOST_CHECK(pso_gen{0u}.evolve(pop).get_x()[0] == pop.get_x()[0]);

    // We check that evolution is deterministic if the
    // seed is controlled and for all algorithmic variants
    // 1) for deterministic optimization
    for (unsigned variant = 1u; variant <= 6u; ++variant) {
        for (unsigned neighb_type = 1u; neighb_type <= 4u; ++neighb_type) {
            problem prob{rosenbrock{10u}};
            population pop1{prob, 5u, 23u};
            pso_gen user_algo1{10u, 0.79, 2., 2., 0.1, variant, neighb_type, 4u, false, 23u};
            user_algo1.set_verbosity(1u);
            pop1 = user_algo1.evolve(pop1);

            population pop2{prob, 5u, 23u};
            pso_gen user_algo2{10u, 0.79, 2., 2., 0.1, variant, neighb_type, 4u, false, 23u};
            user_algo2.set_verbosity(1u);
            pop2 = user_algo2.evolve(pop2);
            BOOST_CHECK(user_algo1.get_log() == user_algo2.get_log());

            population pop3{prob, 5u, 23u};
            user_algo2.set_seed(23u);
            pop3 = user_algo2.evolve(pop3);
            BOOST_CHECK(user_algo1.get_log() == user_algo2.get_log());
        }
    }
    // And with active memory
    for (unsigned variant = 1u; variant <= 6u; ++variant) {
        for (unsigned neighb_type = 1u; neighb_type <= 4u; ++neighb_type) {
            problem prob{rosenbrock{10u}};
            population pop1{prob, 5u, 23u};
            pso_gen user_algo1{10u, 0.79, 2., 2., 0.1, variant, neighb_type, 4u, true, 23u};
            user_algo1.set_verbosity(1u);
            pop1 = user_algo1.evolve(pop1);

            population pop2{prob, 5u, 23u};
            pso_gen user_algo2{10u, 0.79, 2., 2., 0.1, variant, neighb_type, 4u, true, 23u};
            user_algo2.set_verbosity(1u);
            pop2 = user_algo2.evolve(pop2);
            BOOST_CHECK(user_algo1.get_log() == user_algo2.get_log());

            population pop3{prob, 5u, 23u};
            pso_gen user_algo3{10u, 0.79, 2., 2., 0.1, variant, neighb_type, 4u, true, 0u};
            user_algo3.set_verbosity(1u);
            user_algo3.set_seed(23u);
            pop3 = user_algo3.evolve(pop3);
            BOOST_CHECK(user_algo1.get_log() == user_algo3.get_log());
        }
    }
    // 2) for stochastic optimization
    for (unsigned variant = 1u; variant <= 6u; ++variant) {
        for (unsigned neighb_type = 1u; neighb_type <= 4u; ++neighb_type) {
            problem prob{my_sto_prob{10u}};
            population pop1{prob, 5u, 23u};
            pso_gen user_algo1{10u, 0.79, 2., 2., 0.1, variant, neighb_type, 4u, false, 23u};
            user_algo1.set_verbosity(1u);
            pop1 = user_algo1.evolve(pop1);

            population pop2{prob, 5u, 23u};
            pso_gen user_algo2{10u, 0.79, 2., 2., 0.1, variant, neighb_type, 4u, false, 23u};
            user_algo2.set_verbosity(1u);
            pop2 = user_algo2.evolve(pop2);
            BOOST_CHECK(user_algo1.get_log() == user_algo2.get_log());

            population pop3{prob, 5u, 23u};
            user_algo2.set_seed(23u);
            pop3 = user_algo2.evolve(pop3);
            BOOST_CHECK(user_algo1.get_log() == user_algo2.get_log());
        }
    }
    // And with active memory
    for (unsigned variant = 1u; variant <= 6u; ++variant) {
        for (unsigned neighb_type = 1u; neighb_type <= 4u; ++neighb_type) {
            problem prob{my_sto_prob{10u}};
            population pop1{prob, 5u, 23u};
            pso_gen user_algo1{10u, 0.79, 2., 2., 0.1, variant, neighb_type, 4u, true, 23u};
            user_algo1.set_verbosity(1u);
            pop1 = user_algo1.evolve(pop1);

            population pop2{prob, 5u, 23u};
            pso_gen user_algo2{10u, 0.79, 2., 2., 0.1, variant, neighb_type, 4u, true, 23u};
            user_algo2.set_verbosity(1u);
            pop2 = user_algo2.evolve(pop2);
            BOOST_CHECK(user_algo1.get_log() == user_algo2.get_log());

            population pop3{prob, 5u, 23u};
            pso_gen user_algo3{10u, 0.79, 2., 2., 0.1, variant, neighb_type, 4u, true, 0u};
            user_algo3.set_verbosity(1u);
            user_algo3.set_seed(23u);
            pop3 = user_algo3.evolve(pop3);
            BOOST_CHECK(user_algo1.get_log() == user_algo3.get_log());
        }
    }
}
BOOST_AUTO_TEST_CASE(setters_getters_test)
{
    pso_gen user_algo{5000u, 0.79, 2., 2., 0.1, 5u, 2u, 4u, false, 23u};
    user_algo.set_verbosity(200u);
    BOOST_CHECK(user_algo.get_verbosity() == 200u);
    user_algo.set_seed(23u);
    BOOST_CHECK(user_algo.get_seed() == 23u);
    BOOST_CHECK(user_algo.get_name().find("Particle Swarm") != std::string::npos);
    BOOST_CHECK(user_algo.get_name().find("GPSO") != std::string::npos);
    BOOST_CHECK(user_algo.get_extra_info().find("Verbosity") != std::string::npos);
    BOOST_CHECK_NO_THROW(user_algo.get_log());
}

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

    // Store the string representation of p.
    std::stringstream ss;
    auto before_text = boost::lexical_cast<std::string>(algo);
    auto before_log = algo.extract<pso_gen>()->get_log();
    // Now serialize, deserialize and compare the result.
    {
        boost::archive::binary_oarchive oarchive(ss);
        oarchive << algo;
    }
    // Change the content of p before deserializing.
    algo = algorithm{};
    {
        boost::archive::binary_iarchive iarchive(ss);
        iarchive >> algo;
    }
    auto after_text = boost::lexical_cast<std::string>(algo);
    auto after_log = algo.extract<pso_gen>()->get_log();
    BOOST_CHECK_EQUAL(before_text, after_text);
    BOOST_CHECK(before_log == after_log);
    for (auto i = 0u; i < before_log.size(); ++i) {
        BOOST_CHECK_EQUAL(std::get<0>(before_log[i]), std::get<0>(after_log[i]));
        BOOST_CHECK_EQUAL(std::get<1>(before_log[i]), std::get<1>(after_log[i]));
        BOOST_CHECK_CLOSE(std::get<2>(before_log[i]), std::get<2>(after_log[i]), 1e-8);
        BOOST_CHECK_CLOSE(std::get<3>(before_log[i]), std::get<3>(after_log[i]), 1e-8);
        BOOST_CHECK_CLOSE(std::get<4>(before_log[i]), std::get<4>(after_log[i]), 1e-8);
        BOOST_CHECK_CLOSE(std::get<5>(before_log[i]), std::get<5>(after_log[i]), 1e-8);
    }
}

BOOST_AUTO_TEST_CASE(bug)
{
    problem prob{rosenbrock{10u}};
    population pop1{prob, 11u, 23u};
    pso_gen user_algo1{10u, 0.79, 2., 2., 0.1, 1u, 1u, 4u, false, 23u};
    user_algo1.set_verbosity(1u);
    pop1 = user_algo1.evolve(pop1);
}

BOOST_AUTO_TEST_CASE(bfe_usage_test_not_stoch)
{
    population pop{rosenbrock{10u}, 30u, 23u};
    pso_gen uda{40u, 0.79, 2., 2., 0.1, 5u, 2u, 4u, false, 23u};
    uda.set_verbosity(1u);
    uda.set_seed(23u);
    uda.set_bfe(bfe{}); // This will use the default bfe.
    pop = uda.evolve(pop);

    population pop_2{rosenbrock{10u}, 30u, 23u};
    pso_gen uda_2{40u, 0.79, 2., 2., 0.1, 5u, 2u, 4u, false, 23u};
    uda_2.set_verbosity(1u);
    uda_2.set_seed(23u);
    pop_2 = uda_2.evolve(pop_2);

    BOOST_CHECK(pop.get_f() == pop_2.get_f());
}

BOOST_AUTO_TEST_CASE(bfe_usage_test_stoch)
{
    population pop{my_sto_prob{25u, 10, 5}, 30u, 23u};
    pso_gen uda{45u, 0.79, 2., 2., 0.1, 5u, 2u, 4u, false, 23u};
    uda.set_verbosity(1u);
    uda.set_seed(23u);
    uda.set_bfe(bfe{}); // This will use the default bfe.
    pop = uda.evolve(pop);

    population pop_2{my_sto_prob{25u, 10, 5}, 30u, 23u};
    pso_gen uda_2{45u, 0.79, 2., 2., 0.1, 5u, 2u, 4u, false, 23u};
    uda_2.set_verbosity(1u);
    uda_2.set_seed(23u);
    pop_2 = uda_2.evolve(pop_2);

    BOOST_CHECK(pop.get_f() == pop_2.get_f());
}