bridgestan_err_test.cpp

#include <stan/services/sample/hmc_nuts_diag_e_adapt.hpp>
#include <gtest/gtest.h>
#include <stan/io/empty_var_context.hpp>
#include <test/test-models/good/tester.hpp>
#include <test/unit/services/instrumented_callbacks.hpp>
#include <iostream>

class ServicesSampleHmcNutsDiagEAdapt : public testing::Test {
 public:
  ServicesSampleHmcNutsDiagEAdapt() : model(context, 0, &model_log) {}

  std::stringstream model_log;
  stan::test::unit::instrumented_logger logger;
  stan::test::unit::instrumented_writer init, parameter, diagnostic;
  stan::io::empty_var_context context;
  stan_model model;
};

template <typename M>
void log_density_gradient(M&& model, bool propto, bool jacobian,
                                     const double* theta_unc, double* val,
                                     double* grad) {
  static thread_local stan::math::ChainableStack thread_instance;
  auto logp = [&model](auto& x) {
    return model.template log_prob<true, true, stan::math::var>(x);
  };
  int N = 5;
  Eigen::VectorXd params_unc = Eigen::VectorXd::Map(theta_unc, N);
  Eigen::VectorXd grad_vec(N);
  stan::math::gradient(logp, params_unc, val, grad, grad + N);
  Eigen::VectorXd::Map(grad, N) = grad_vec;
}
TEST_F(ServicesSampleHmcNutsDiagEAdapt, call_count) {
  int param_unc_num_ = 5;
  int N = param_unc_num_;
  double val = 0;
  double theta_unc[5];
  theta_unc[0] = 1;
  theta_unc[1] = 2;
  theta_unc[2] = 3;
  theta_unc[3] = 4;
  theta_unc[4] = 5;
  double grad_vs[5];
  Eigen::VectorXd params_unc = Eigen::VectorXd::Map(theta_unc, N);
  log_density_gradient(model, true, true, theta_unc, &val, grad_vs);
  Eigen::Map<Eigen::VectorXd> grad_vec = Eigen::Map<Eigen::VectorXd>(grad_vs, N);
  std::cout << "val: " << val << std::endl;
  std::cout << "\ngrad:\n" << grad_vec << std::endl;
}