TEST(McmcNutsBaseNuts, transition) {
rng_t base_rng(0);
int model_size = 1;
double init_momentum = 1.5;
stan::mcmc::ps_point z_init(model_size);
z_init.q(0) = 0;
z_init.p(0) = init_momentum;
stan::mcmc::mock_model model(model_size);
stan::mcmc::mock_nuts sampler(model, base_rng);
sampler.set_nominal_stepsize(1);
sampler.set_stepsize_jitter(0);
sampler.sample_stepsize();
sampler.z() = z_init;
std::stringstream output_stream;
stan::interface_callbacks::writer::stream_writer writer(output_stream);
std::stringstream error_stream;
stan::interface_callbacks::writer::stream_writer error_writer(error_stream);
stan::mcmc::sample init_sample(z_init.q, 0, 0);
stan::mcmc::sample s = sampler.transition(init_sample, writer, error_writer);
EXPECT_EQ(31.5, s.cont_params()(0));
EXPECT_EQ(0, s.log_prob());
EXPECT_EQ(1, s.accept_stat());
EXPECT_EQ("", output_stream.str());
EXPECT_EQ("", error_stream.str());
}
TEST(McmcBaseStaticHMC, set_nominal_stepsize_and_T) {
rng_t base_rng(0);
std::vector<double> q(5, 1.0);
std::vector<int> r(2, 2);
stan::mcmc::mock_model model(q.size());
stan::mcmc::mock_static_hmc sampler(model, base_rng, &std::cout, &std::cerr);
double old_epsilon = 1.0;
double old_T = 3.0;
sampler.set_nominal_stepsize_and_T(old_epsilon, old_T);
EXPECT_EQ(old_epsilon, sampler.get_nominal_stepsize());
EXPECT_EQ(old_T, sampler.get_T());
EXPECT_EQ(true, sampler.get_L() > 0);
sampler.set_nominal_stepsize_and_T(-0.1, 5.0);
EXPECT_EQ(old_epsilon, sampler.get_nominal_stepsize());
EXPECT_EQ(old_T, sampler.get_T());
sampler.set_nominal_stepsize_and_T(5.0, -0.1);
EXPECT_EQ(old_epsilon, sampler.get_nominal_stepsize());
EXPECT_EQ(old_T, sampler.get_T());
}
TEST(McmcDenseEMetric, streams) {
stan::test::capture_std_streams();
rng_t base_rng(0);
Eigen::VectorXd q(2);
q(0) = 5;
q(1) = 1;
stan::mcmc::mock_model model(q.size());
// typedef to use within Google Test macros
typedef stan::mcmc::dense_e_metric<stan::mcmc::mock_model, rng_t> dense_e;
EXPECT_NO_THROW(dense_e metric(model, 0));
std::stringstream metric_output;
EXPECT_NO_THROW(dense_e metric(model, &metric_output));
EXPECT_EQ("", metric_output.str());
stan::test::reset_std_streams();
EXPECT_EQ("", stan::test::cout_ss.str());
EXPECT_EQ("", stan::test::cerr_ss.str());
}
TEST(Services, do_bfgs_optimize__lbfgs) {
std::vector<double> cont_vector(2);
cont_vector[0] = -1; cont_vector[1] = 1;
std::vector<int> disc_vector;
static const std::string DATA("");
std::stringstream data_stream(DATA);
stan::io::dump dummy_context(data_stream);
Model model(dummy_context);
typedef stan::optimization::BFGSLineSearch<Model,stan::optimization::LBFGSUpdate<> > Optimizer_LBFGS;
Optimizer_LBFGS lbfgs(model, cont_vector, disc_vector, &std::cout);
double lp = 0;
bool save_iterations = true;
int refresh = 0;
int return_code;
unsigned int random_seed = 0;
rng_t base_rng(random_seed);
std::fstream* output_stream = 0;
mock_callback callback;
return_code = stan::services::optimization::do_bfgs_optimize(model, lbfgs, base_rng,
lp, cont_vector, disc_vector,
output_stream, &std::cout,
save_iterations, refresh,
callback);
EXPECT_FLOAT_EQ(return_code, 0);
EXPECT_EQ(35, callback.n);
}
TEST(McmcDerivedNuts, compute_criterion_dense_e) {
rng_t base_rng(0);
int model_size = 1;
stan::mcmc::ps_point start(model_size);
stan::mcmc::dense_e_point finish(model_size);
Eigen::VectorXd rho(model_size);
stan::mcmc::mock_model model(model_size);
stan::mcmc::dense_e_nuts<stan::mcmc::mock_model, rng_t> sampler(model, base_rng, 0, 0);
start.q(0) = 1;
start.p(0) = 1;
finish.q(0) = 2;
finish.p(0) = 1;
rho = start.p + finish.p;
EXPECT_EQ(true, sampler.compute_criterion(start, finish, rho));
start.q(0) = 1;
start.p(0) = 1;
finish.q(0) = 2;
finish.p(0) = -1;
rho = start.p + finish.p;
EXPECT_FALSE(sampler.compute_criterion(start, finish, rho));
}
TEST(McmcStaticBaseStaticHMC, set_nominal_stepsize_and_L) {
rng_t base_rng(0);
std::vector<double> q(5, 1.0);
std::vector<int> r(2, 2);
stan::mcmc::mock_model model(q.size());
stan::mcmc::mock_static_hmc sampler(model, base_rng);
double old_epsilon = 1.0;
int old_L = 10;
sampler.set_nominal_stepsize_and_L(old_epsilon, old_L);
EXPECT_EQ(old_epsilon, sampler.get_nominal_stepsize());
EXPECT_EQ(old_L, sampler.get_L());
EXPECT_EQ(true, sampler.get_T() > 0);
sampler.set_nominal_stepsize_and_L(-0.1, 5);
EXPECT_EQ(old_epsilon, sampler.get_nominal_stepsize());
EXPECT_EQ(old_L, sampler.get_L());
sampler.set_nominal_stepsize_and_T(5.0, -1);
EXPECT_EQ(old_epsilon, sampler.get_nominal_stepsize());
EXPECT_EQ(old_L, sampler.get_L());
}
TEST(McmcUnitEMetric, sample_p) {
rng_t base_rng(0);
std::vector<double> q(5, 1.0);
std::vector<int> r(2, 2);
stan::mcmc::mock_model model(q.size());
stan::mcmc::unit_e_metric<stan::mcmc::mock_model, rng_t> metric(model, &std::cout);
stan::mcmc::unit_e_point z(q.size(), r.size());
int n_samples = 1000;
double m = 0;
double m2 = 0;
for (int i = 0; i < n_samples; ++i) {
metric.sample_p(z, base_rng);
double T = metric.T(z);
double delta = T - m;
m += delta / static_cast<double>(i + 1);
m2 += delta * (T - m);
}
double var = m2 / (n_samples + 1.0);
// Mean within 5sigma of expected value (d / 2)
EXPECT_EQ(true, fabs(m - 0.5 * q.size()) < 5.0 * sqrt(var));
// Variance within 10% of expected value (d / 2)
EXPECT_EQ(true, fabs(var - 0.5 * q.size()) < 0.1 * q.size());
}
TEST(advi_test, hier_logistic_cp_constraint_meanfield) {
// Create mock data_var_context
std::fstream data_stream("src/test/test-models/good/variational/hier_logistic.data.R",
std::fstream::in);
stan::io::dump data_var_context(data_stream);
data_stream.close();
std::stringstream output;
output.clear();
// Instantiate model
Model_cp my_model(data_var_context);
// RNG
rng_t base_rng(0);
// Dummy input
Eigen::VectorXd cont_params = Eigen::VectorXd::Zero(my_model.num_params_r());
// ADVI
stan::variational::advi<Model_cp, stan::variational::normal_meanfield, rng_t> test_advi(my_model,
cont_params,
base_rng,
10,
100,
100,
1);
stan::interface_callbacks::writer::noop_writer writer;
test_advi.run(0.01, false, 50, 1, 2e4,
writer, writer, writer);
}
TEST(McmcNuts, instantiaton_test) {
rng_t base_rng(4839294);
std::stringstream output;
stan::callbacks::stream_writer writer(output);
std::stringstream error_stream;
stan::callbacks::stream_writer error_writer(error_stream);
std::fstream empty_stream("", std::fstream::in);
stan::io::dump data_var_context(empty_stream);
gauss3D_model_namespace::gauss3D_model model(data_var_context);
stan::mcmc::unit_e_nuts<gauss3D_model_namespace::gauss3D_model, rng_t>
unit_e_sampler(model, base_rng);
stan::mcmc::diag_e_nuts<gauss3D_model_namespace::gauss3D_model, rng_t>
diag_e_sampler(model, base_rng);
stan::mcmc::dense_e_nuts<gauss3D_model_namespace::gauss3D_model, rng_t>
dense_e_sampler(model, base_rng);
stan::mcmc::adapt_unit_e_nuts<gauss3D_model_namespace::gauss3D_model, rng_t>
adapt_unit_e_sampler(model, base_rng);
stan::mcmc::adapt_diag_e_nuts<gauss3D_model_namespace::gauss3D_model, rng_t>
adapt_diag_e_sampler(model, base_rng);
stan::mcmc::adapt_dense_e_nuts<gauss3D_model_namespace::gauss3D_model, rng_t>
adapt_dense_e_sampler(model, base_rng);
}
TEST(McmcDiagEMetric, sample_p) {
rng_t base_rng(0);
Eigen::VectorXd q(2);
q(0) = 5;
q(1) = 1;
stan::mcmc::mock_model model(q.size());
stan::mcmc::diag_e_metric<stan::mcmc::mock_model, rng_t> metric(model);
stan::mcmc::diag_e_point z(q.size());
int n_samples = 1000;
double m = 0;
double m2 = 0;
for (int i = 0; i < n_samples; ++i) {
metric.sample_p(z, base_rng);
double tau = metric.tau(z);
double delta = tau - m;
m += delta / static_cast<double>(i + 1);
m2 += delta * (tau - m);
}
double var = m2 / (n_samples + 1.0);
// Mean within 5sigma of expected value (d / 2)
EXPECT_TRUE(std::fabs(m - 0.5 * q.size()) < 5.0 * sqrt(var));
// Variance within 10% of expected value (d / 2)
EXPECT_TRUE(std::fabs(var - 0.5 * q.size()) < 0.1 * q.size());
}
TEST(McmcBaseStaticHMC, set_nominal_stepsize) {
rng_t base_rng(0);
std::vector<double> q(5, 1.0);
std::vector<int> r(2, 2);
stan::mcmc::mock_model model(q.size());
std::stringstream output, error;
stan::mcmc::mock_static_hmc sampler(model, base_rng, &output, &error);
double old_epsilon = 1.0;
sampler.set_nominal_stepsize(old_epsilon);
EXPECT_EQ(old_epsilon, sampler.get_nominal_stepsize());
EXPECT_EQ(true, sampler.get_L() > 0);
sampler.set_nominal_stepsize(-0.1);
EXPECT_EQ(old_epsilon, sampler.get_nominal_stepsize());
EXPECT_EQ("", output.str());
EXPECT_EQ("", error.str());
}
TEST(McmcNutsBaseNuts, build_tree_test) {
rng_t base_rng(0);
int model_size = 1;
double init_momentum = 1.5;
stan::mcmc::ps_point z_init(model_size);
z_init.q(0) = 0;
z_init.p(0) = init_momentum;
stan::mcmc::ps_point z_propose(model_size);
Eigen::VectorXd p_sharp_left = Eigen::VectorXd::Zero(model_size);
Eigen::VectorXd p_sharp_right = Eigen::VectorXd::Zero(model_size);
Eigen::VectorXd rho = z_init.p;
double log_sum_weight = -std::numeric_limits<double>::infinity();
double H0 = -0.1;
int n_leapfrog = 0;
double sum_metro_prob = 0;
stan::mcmc::mock_model model(model_size);
stan::mcmc::mock_nuts sampler(model, base_rng);
sampler.set_nominal_stepsize(1);
sampler.set_stepsize_jitter(0);
sampler.sample_stepsize();
sampler.z() = z_init;
std::stringstream debug, info, warn, error, fatal;
stan::callbacks::stream_logger logger(debug, info, warn, error, fatal);
bool valid_subtree = sampler.build_tree(3, z_propose,
p_sharp_left, p_sharp_right, rho,
H0, 1, n_leapfrog, log_sum_weight,
sum_metro_prob, logger);
EXPECT_TRUE(valid_subtree);
EXPECT_EQ(init_momentum * (n_leapfrog + 1), rho(0));
EXPECT_EQ(1, p_sharp_left(0));
EXPECT_EQ(1, p_sharp_right(0));
EXPECT_EQ(8 * init_momentum, sampler.z().q(0));
EXPECT_EQ(init_momentum, sampler.z().p(0));
EXPECT_EQ(8, n_leapfrog);
EXPECT_FLOAT_EQ(H0 + std::log(n_leapfrog), log_sum_weight);
EXPECT_FLOAT_EQ(std::exp(H0) * n_leapfrog, sum_metro_prob);
EXPECT_EQ("", debug.str());
EXPECT_EQ("", info.str());
EXPECT_EQ("", warn.str());
EXPECT_EQ("", error.str());
EXPECT_EQ("", fatal.str());
}
TEST(StanIoMcmcWriter, write_diagnostic_names) {
// Model
std::fstream data_stream("", std::fstream::in);
stan::io::dump data_var_context(data_stream);
data_stream.close();
std::stringstream output;
io_example_model_namespace::io_example_model model(data_var_context, &output);
// Sample
Eigen::VectorXd real(2);
real(0) = 1.43;
real(1) = 2.71;
double log_prob = 3.14;
double accept_stat = 0.84;
stan::mcmc::sample sample(real, log_prob, accept_stat);
// Sampler
typedef boost::ecuyer1988 rng_t;
rng_t base_rng(0);
stan::mcmc::adapt_diag_e_nuts<io_example_model_namespace::io_example_model, rng_t>
sampler(model, base_rng, 0, 0);
sampler.seed(real);
// Writer
std::stringstream sample_stream;
std::stringstream diagnostic_stream;
std::stringstream message_stream;
stan::interface::recorder::csv sample_recorder(&sample_stream, "# ");
stan::interface::recorder::csv diagnostic_recorder(&diagnostic_stream, "# ");
stan::interface::recorder::messages message_recorder(&message_stream, "# ");
stan::io::mcmc_writer<io_example_model_namespace::io_example_model,
stan::interface::recorder::csv,
stan::interface::recorder::csv,
stan::interface::recorder::messages>
writer(sample_recorder, diagnostic_recorder, message_recorder);
writer.write_diagnostic_names(sample, &sampler, model);
std::string line;
std::getline(diagnostic_stream, line);
// FIXME: make this work, too
EXPECT_EQ("lp__,accept_stat__,stepsize__,treedepth__,n_leapfrog__,n_divergent__,mu1,mu2,p_mu1,p_mu2,g_mu1,g_mu2", line);
EXPECT_EQ("", message_stream.str());
EXPECT_EQ("", output.str());
}
TEST(McmcBaseNuts, slice_criterion) {
rng_t base_rng(0);
int model_size = 1;
double init_momentum = 1.5;
Eigen::VectorXd rho = Eigen::VectorXd::Zero(model_size);
stan::mcmc::ps_point z_init(model_size);
z_init.q(0) = 0;
z_init.p(0) = init_momentum;
stan::mcmc::ps_point z_propose(model_size);
stan::mcmc::nuts_util util;
util.log_u = 0;
util.H0 = 0;
util.sign = 1;
util.n_tree = 0;
util.sum_prob = 0;
std::stringstream output, error;
stan::mcmc::mock_model model(model_size);
stan::mcmc::divergent_nuts sampler(model, base_rng, &output, &error);
sampler.set_nominal_stepsize(1);
sampler.set_stepsize_jitter(0);
sampler.sample_stepsize();
sampler.z() = z_init;
int n_valid = 0;
sampler.z().V = -750;
n_valid = sampler.build_tree(0, rho, &z_init, z_propose, util);
EXPECT_EQ(1, n_valid);
EXPECT_EQ(0, sampler.n_divergent_);
sampler.z().V = -250;
n_valid = sampler.build_tree(0, rho, &z_init, z_propose, util);
EXPECT_EQ(0, n_valid);
EXPECT_EQ(0, sampler.n_divergent_);
sampler.z().V = 750;
n_valid = sampler.build_tree(0, rho, &z_init, z_propose, util);
EXPECT_EQ(0, n_valid);
EXPECT_EQ(1, sampler.n_divergent_);
EXPECT_EQ("", output.str());
EXPECT_EQ("", error.str());
}
TEST(McmcNutsBaseNuts, build_tree) {
rng_t base_rng(0);
int model_size = 1;
double init_momentum = 1.5;
stan::mcmc::ps_point z_init(model_size);
z_init.q(0) = 0;
z_init.p(0) = init_momentum;
stan::mcmc::ps_point z_propose(model_size);
Eigen::VectorXd rho = z_init.p;
double log_sum_weight = -std::numeric_limits<double>::infinity();
double H0 = -0.1;
int n_leapfrog = 0;
double sum_metro_prob = 0;
stan::mcmc::mock_model model(model_size);
stan::mcmc::mock_nuts sampler(model, base_rng);
sampler.set_nominal_stepsize(1);
sampler.set_stepsize_jitter(0);
sampler.sample_stepsize();
sampler.z() = z_init;
std::stringstream output;
stan::interface_callbacks::writer::stream_writer writer(output);
std::stringstream error_stream;
stan::interface_callbacks::writer::stream_writer error_writer(error_stream);
bool valid_subtree = sampler.build_tree(3, rho, z_propose,
H0, 1, n_leapfrog, log_sum_weight,
sum_metro_prob, writer, error_writer);
EXPECT_TRUE(valid_subtree);
EXPECT_EQ(init_momentum * (n_leapfrog + 1), rho(0));
EXPECT_EQ(8 * init_momentum, sampler.z().q(0));
EXPECT_EQ(init_momentum, sampler.z().p(0));
EXPECT_EQ(8, n_leapfrog);
EXPECT_FLOAT_EQ(H0 + std::log(n_leapfrog), log_sum_weight);
EXPECT_FLOAT_EQ(std::exp(H0) * n_leapfrog, sum_metro_prob);
EXPECT_EQ("", output.str());
EXPECT_EQ("", error_stream.str());
}
TEST(McmcNutsBaseNuts, set_max_delta_test) {
rng_t base_rng(0);
Eigen::VectorXd q(2);
q(0) = 5;
q(1) = 1;
stan::mcmc::mock_model model(q.size());
stan::mcmc::mock_nuts sampler(model, base_rng);
double old_max_delta = 10;
sampler.set_max_delta(old_max_delta);
EXPECT_EQ(old_max_delta, sampler.get_max_delta());
}
TEST(McmcHmcIntegratorsImplLeapfrog, unit_e_energy_conservation) {
rng_t base_rng(0);
std::fstream data_stream(std::string("").c_str(), std::fstream::in);
stan::io::dump data_var_context(data_stream);
data_stream.close();
std::stringstream model_output;
std::stringstream debug, info, warn, error, fatal;
stan::callbacks::stream_logger logger(debug, info, warn, error, fatal);
gauss_model_namespace::gauss_model model(data_var_context, &model_output);
stan::mcmc::impl_leapfrog<
stan::mcmc::unit_e_metric<gauss_model_namespace::gauss_model, rng_t> >
integrator;
stan::mcmc::unit_e_metric<gauss_model_namespace::gauss_model, rng_t> metric(model);
stan::mcmc::unit_e_point z(1);
z.q(0) = 1;
z.p(0) = 1;
metric.init(z, logger);
double H0 = metric.H(z);
double aveDeltaH = 0;
double epsilon = 1e-3;
double tau = 6.28318530717959;
size_t L = tau / epsilon;
for (size_t n = 0; n < L; ++n) {
integrator.evolve(z, metric, epsilon, logger);
double deltaH = metric.H(z) - H0;
aveDeltaH += (deltaH - aveDeltaH) / double(n + 1);
}
// Average error in Hamiltonian should be O(epsilon^{2})
// in general, smaller for the gauss_model in this case due to cancellations
EXPECT_NEAR(aveDeltaH, 0, epsilon * epsilon);
EXPECT_EQ("", model_output.str());
EXPECT_EQ("", debug.str());
EXPECT_EQ("", info.str());
EXPECT_EQ("", warn.str());
EXPECT_EQ("", error.str());
EXPECT_EQ("", fatal.str());
}
TEST(McmcBaseNuts, build_tree) {
rng_t base_rng(0);
int model_size = 1;
double init_momentum = 1.5;
Eigen::VectorXd rho = Eigen::VectorXd::Zero(model_size);
stan::mcmc::ps_point z_init(model_size);
z_init.q(0) = 0;
z_init.p(0) = init_momentum;
stan::mcmc::ps_point z_propose(model_size);
stan::mcmc::nuts_util util;
util.log_u = -1;
util.H0 = -0.1;
util.sign = 1;
util.n_tree = 0;
util.sum_prob = 0;
std::stringstream output, error;
stan::mcmc::mock_model model(model_size);
stan::mcmc::mock_nuts sampler(model, base_rng, &output, &error);
sampler.set_nominal_stepsize(1);
sampler.set_stepsize_jitter(0);
sampler.sample_stepsize();
sampler.z() = z_init;
int n_valid = sampler.build_tree(3, rho, &z_init, z_propose, util);
EXPECT_EQ(8, n_valid);
EXPECT_EQ(8, util.n_tree);
EXPECT_FLOAT_EQ(std::exp(util.H0) * util.n_tree, util.sum_prob);
EXPECT_EQ(init_momentum * util.n_tree, rho(0));
EXPECT_EQ(init_momentum, z_init.q(0));
EXPECT_EQ(init_momentum, z_init.p(0));
EXPECT_EQ(8 * init_momentum, sampler.z().q(0));
EXPECT_EQ(init_momentum, sampler.z().p(0));
EXPECT_EQ("", output.str());
EXPECT_EQ("", error.str());
}
TEST(McmcBaseHMC, point_construction) {
rng_t base_rng(0);
std::vector<double> q(5, 1.0);
std::vector<int> r(2, 2);
stan::mcmc::mock_model model(q.size());
stan::mcmc::mock_hmc sampler(model, base_rng, &std::cout, &std::cerr);
EXPECT_EQ(q.size(), sampler.z().q.size());
EXPECT_EQ(q.size(), sampler.z().g.size());
}
TEST(McmcNutsBaseNuts, rho_aggregation_test) {
rng_t base_rng(0);
int model_size = 1;
double init_momentum = 1.5;
stan::mcmc::ps_point z_init(model_size);
z_init.q(0) = 0;
z_init.p(0) = init_momentum;
stan::mcmc::ps_point z_propose(model_size);
Eigen::VectorXd p_sharp_left = Eigen::VectorXd::Zero(model_size);
Eigen::VectorXd p_sharp_right = Eigen::VectorXd::Zero(model_size);
Eigen::VectorXd rho = z_init.p;
double log_sum_weight = -std::numeric_limits<double>::infinity();
double H0 = -0.1;
int n_leapfrog = 0;
double sum_metro_prob = 0;
stan::mcmc::mock_model model(model_size);
stan::mcmc::rho_inspector_mock_nuts sampler(model, base_rng);
sampler.set_nominal_stepsize(1);
sampler.set_stepsize_jitter(0);
sampler.sample_stepsize();
sampler.z() = z_init;
std::stringstream debug, info, warn, error, fatal;
stan::callbacks::stream_logger logger(debug, info, warn, error, fatal);
sampler.build_tree(3, z_propose,
p_sharp_left, p_sharp_right, rho,
H0, 1, n_leapfrog, log_sum_weight,
sum_metro_prob, logger);
EXPECT_EQ(7, sampler.rho_values.size());
EXPECT_EQ(2 * init_momentum, sampler.rho_values.at(0));
EXPECT_EQ(2 * init_momentum, sampler.rho_values.at(1));
EXPECT_EQ(4 * init_momentum, sampler.rho_values.at(2));
EXPECT_EQ(2 * init_momentum, sampler.rho_values.at(3));
EXPECT_EQ(2 * init_momentum, sampler.rho_values.at(4));
EXPECT_EQ(4 * init_momentum, sampler.rho_values.at(5));
EXPECT_EQ(8 * init_momentum, sampler.rho_values.at(6));
}
TEST(McmcUnitENuts, transition_test) {
rng_t base_rng(4839294);
stan::mcmc::unit_e_point z_init(3);
z_init.q(0) = 1;
z_init.q(1) = -1;
z_init.q(2) = 1;
z_init.p(0) = -1;
z_init.p(1) = 1;
z_init.p(2) = -1;
std::stringstream output_stream;
stan::interface_callbacks::writer::stream_writer writer(output_stream);
std::stringstream error_stream;
stan::interface_callbacks::writer::stream_writer error_writer(error_stream);
std::fstream empty_stream("", std::fstream::in);
stan::io::dump data_var_context(empty_stream);
gauss3D_model_namespace::gauss3D_model model(data_var_context);
stan::mcmc::unit_e_nuts<gauss3D_model_namespace::gauss3D_model, rng_t>
sampler(model, base_rng);
sampler.z() = z_init;
sampler.init_hamiltonian(writer, error_writer);
sampler.set_nominal_stepsize(0.1);
sampler.set_stepsize_jitter(0);
sampler.sample_stepsize();
stan::mcmc::sample init_sample(z_init.q, 0, 0);
stan::mcmc::sample s = sampler.transition(init_sample, writer, error_writer);
EXPECT_EQ(4, sampler.depth_);
EXPECT_EQ((2 << 3) - 1, sampler.n_leapfrog_);
EXPECT_FALSE(sampler.divergent_);
EXPECT_FLOAT_EQ(1.8718261, s.cont_params()(0));
EXPECT_FLOAT_EQ(-0.74208695, s.cont_params()(1));
EXPECT_FLOAT_EQ( 1.5202962, s.cont_params()(2));
EXPECT_FLOAT_EQ(-3.1828632, s.log_prob());
EXPECT_FLOAT_EQ(0.99629009, s.accept_stat());
EXPECT_EQ("", output_stream.str());
EXPECT_EQ("", error_stream.str());
}
TEST(McmcDenseEMetric, sample_p) {
rng_t base_rng(0);
Eigen::Matrix2d m(2,2);
m(0, 0) = 3.0;
m(1, 0) = -2.0;
m(0, 1) = -2.0;
m(1, 1) = 4.0;
Eigen::Matrix2d m_inv = m.inverse();
stan::mcmc::mock_model model(2);
stan::mcmc::dense_e_metric<stan::mcmc::mock_model, rng_t> metric(model);
stan::mcmc::dense_e_point z(2);
z.set_metric(m_inv);
int n_samples = 1000;
Eigen::Matrix2d sample_cov(2,2);
sample_cov(0, 0) = 0.0;
sample_cov(0, 1) = 0.0;
sample_cov(1, 0) = 0.0;
sample_cov(1, 1) = 0.0;
for (int i = 0; i < n_samples; ++i) {
metric.sample_p(z, base_rng);
sample_cov(0, 0) += z.p[0] * z.p[0] / n_samples;
sample_cov(0, 1) += z.p[0] * z.p[1] / n_samples;
sample_cov(1, 0) += z.p[1] * z.p[0] / n_samples;
sample_cov(1, 1) += z.p[1] * z.p[1] / n_samples;
}
Eigen::Matrix2d var(2,2);
var(0, 0) = 2 * m(0, 0);
var(1, 0) = m(1, 0) * m(1, 0) + m(1, 1) * m(0, 0);
var(0, 1) = m(0, 1) * m(0, 1) + m(1, 1) * m(0, 0);
var(1, 1) = 2 * m(1, 1);
// Covariance matrix within 5sigma of expected value (comes from a Wishart distribution)
EXPECT_TRUE(std::fabs(m(0, 0) - sample_cov(0, 0)) < 5.0 * sqrt(var(0, 0) / n_samples));
EXPECT_TRUE(std::fabs(m(1, 0) - sample_cov(1, 0)) < 5.0 * sqrt(var(1, 0) / n_samples));
EXPECT_TRUE(std::fabs(m(0, 1) - sample_cov(0, 1)) < 5.0 * sqrt(var(0, 1) / n_samples));
EXPECT_TRUE(std::fabs(m(1, 1) - sample_cov(1, 1)) < 5.0 * sqrt(var(1, 1) / n_samples));
}
TEST(McmcNutsBaseNuts, set_max_depth) {
rng_t base_rng(0);
Eigen::VectorXd q(2);
q(0) = 5;
q(1) = 1;
stan::mcmc::mock_model model(q.size());
stan::mcmc::mock_nuts sampler(model, base_rng);
int old_max_depth = 1;
sampler.set_max_depth(old_max_depth);
EXPECT_EQ(old_max_depth, sampler.get_max_depth());
sampler.set_max_depth(-1);
EXPECT_EQ(old_max_depth, sampler.get_max_depth());
}
TEST(McmcBaseNuts, set_max_delta) {
rng_t base_rng(0);
Eigen::VectorXd q(2);
q(0) = 5;
q(1) = 1;
std::stringstream output, error;
stan::mcmc::mock_model model(q.size());
stan::mcmc::mock_nuts sampler(model, base_rng, &output, &error);
double old_max_delta = 10;
sampler.set_max_delta(old_max_delta);
EXPECT_EQ(old_max_delta, sampler.get_max_delta());
EXPECT_EQ("", output.str());
EXPECT_EQ("", error.str());
}
TEST(McmcBaseHMC, seed) {
rng_t base_rng(0);
std::vector<double> q(5, 1.0);
std::vector<int> r;
stan::mcmc::mock_model model(q.size());
stan::mcmc::mock_hmc sampler(model, base_rng, &std::cout, &std::cerr);
std::vector<double> q_seed(q.size(), -1.0);
sampler.seed(q_seed, r);
for (int i = 0; i < q.size(); ++i)
EXPECT_EQ(q_seed.at(i), sampler.z().q.at(i));
}
TEST(McmcBaseHMC, set_nominal_stepsize) {
rng_t base_rng(0);
std::vector<double> q(5, 1.0);
std::vector<int> r(2, 2);
stan::mcmc::mock_model model(q.size());
stan::mcmc::mock_hmc sampler(model, base_rng, &std::cout, &std::cerr);
double old_epsilon = 1.0;
sampler.set_nominal_stepsize(old_epsilon);
EXPECT_EQ(old_epsilon, sampler.get_nominal_stepsize());
sampler.set_nominal_stepsize(-0.1);
EXPECT_EQ(old_epsilon, sampler.get_nominal_stepsize());
}
TEST(McmcHmcIntegratorsExplLeapfrog, energy_conservation) {
rng_t base_rng(0);
std::fstream data_stream(std::string("").c_str(), std::fstream::in);
stan::io::dump data_var_context(data_stream);
data_stream.close();
gauss_namespace::gauss model(data_var_context, &std::cout);
stan::mcmc::expl_leapfrog<
stan::mcmc::unit_e_metric<gauss_namespace::gauss, rng_t>,
stan::mcmc::unit_e_point> integrator;
stan::mcmc::unit_e_metric<gauss_namespace::gauss, rng_t> metric(model, &std::cout);
stan::mcmc::unit_e_point z(1);
z.q(0) = 1;
z.p(0) = 1;
metric.update(z);
double H0 = metric.H(z);
double aveDeltaH = 0;
double epsilon = 1e-3;
double tau = 6.28318530717959;
size_t L = tau / epsilon;
for (size_t n = 0; n < L; ++n) {
integrator.evolve(z, metric, epsilon);
double deltaH = metric.H(z) - H0;
aveDeltaH += (deltaH - aveDeltaH) / double(n + 1);
}
// Average error in Hamiltonian should be O(epsilon^{2})
// in general, smaller for the gaussian case due to cancellations
EXPECT_NEAR(aveDeltaH, 0, epsilon * epsilon);
}
TEST(McmcStaticBaseStaticHMC, set_T) {
rng_t base_rng(0);
std::vector<double> q(5, 1.0);
std::vector<int> r(2, 2);
stan::mcmc::mock_model model(q.size());
stan::mcmc::mock_static_hmc sampler(model, base_rng);
double old_T = 3.0;
sampler.set_T(old_T);
EXPECT_EQ(old_T, sampler.get_T());
EXPECT_EQ(true, sampler.get_L() > 0);
sampler.set_T(-0.1);
EXPECT_EQ(old_T, sampler.get_T());
}
TEST(McmcNutsBaseNuts, transition) {
rng_t base_rng(0);
int model_size = 1;
double init_momentum = 1.5;
stan::mcmc::ps_point z_init(model_size);
z_init.q(0) = 0;
z_init.p(0) = init_momentum;
stan::mcmc::mock_model model(model_size);
stan::mcmc::mock_nuts sampler(model, base_rng);
sampler.set_nominal_stepsize(1);
sampler.set_stepsize_jitter(0);
sampler.sample_stepsize();
sampler.z() = z_init;
std::stringstream debug, info, warn, error, fatal;
stan::callbacks::stream_logger logger(debug, info, warn, error, fatal);
stan::mcmc::sample init_sample(z_init.q, 0, 0);
// Transition will expand trajectory until max_depth is hit
stan::mcmc::sample s = sampler.transition(init_sample, logger);
EXPECT_EQ(sampler.get_max_depth(), sampler.depth_);
EXPECT_EQ((2 << (sampler.get_max_depth() - 1)) - 1, sampler.n_leapfrog_);
EXPECT_FALSE(sampler.divergent_);
EXPECT_EQ(21 * init_momentum, s.cont_params()(0));
EXPECT_EQ(0, s.log_prob());
EXPECT_EQ(1, s.accept_stat());
EXPECT_EQ("", debug.str());
EXPECT_EQ("", info.str());
EXPECT_EQ("", warn.str());
EXPECT_EQ("", error.str());
EXPECT_EQ("", fatal.str());
}
TEST(McmcNutsDerivedNuts, compute_criterion_softabs) {
rng_t base_rng(0);
int model_size = 1;
stan::mcmc::softabs_point start(model_size);
stan::mcmc::softabs_point finish(model_size);
Eigen::VectorXd p_sharp_start(model_size);
Eigen::VectorXd p_sharp_finish(model_size);
Eigen::VectorXd rho(model_size);
stan::mcmc::mock_model model(model_size);
stan::mcmc::softabs_nuts<stan::mcmc::mock_model, rng_t>
sampler(model, base_rng);
start.q(0) = 1;
start.p(0) = 1;
finish.q(0) = 2;
finish.p(0) = 1;
p_sharp_start = start.p;
p_sharp_finish = finish.p;
rho = start.p + finish.p;
EXPECT_TRUE(sampler.compute_criterion(p_sharp_start, p_sharp_finish, rho));
start.q(0) = 1;
start.p(0) = 1;
finish.q(0) = 2;
finish.p(0) = -1;
p_sharp_start = start.p;
p_sharp_finish = finish.p;
rho = start.p + finish.p;
EXPECT_FALSE(sampler.compute_criterion(p_sharp_start, p_sharp_finish, rho));
}
