skew_double_exponential_lcdf.hpp

Go to the documentation of this file.

#ifndef STAN_MATH_OPENCL_PRIM_DOUBLE_SKEW_DOUBLE_EXPONENTIAL_LCDF_HPP
#define STAN_MATH_OPENCL_PRIM_DOUBLE_SKEW_DOUBLE_EXPONENTIAL_LCDF_HPP
#ifdef STAN_OPENCL
 
#include <stan/math/prim/meta.hpp>
#include <stan/math/prim/err.hpp>
#include <stan/math/prim/fun/constants.hpp>
#include <stan/math/prim/fun/elt_divide.hpp>
#include <stan/math/prim/fun/elt_multiply.hpp>
#include <stan/math/opencl/kernel_generator.hpp>
#include <stan/math/prim/functor/partials_propagator.hpp>
 
namespace stan {
namespace math {
 
template <typename T_y_cl, typename T_loc_cl, typename T_scale_cl,
          typename T_skewness_cl,
          require_all_prim_or_rev_kernel_expression_t<
              T_y_cl, T_loc_cl, T_scale_cl, T_skewness_cl>* = nullptr,
          require_any_not_stan_scalar_t<T_y_cl, T_loc_cl, T_scale_cl,
                                        T_skewness_cl>* = nullptr>
inline return_type_t<T_y_cl, T_loc_cl, T_scale_cl, T_skewness_cl>
skew_double_exponential_lcdf(const T_y_cl& y, const T_loc_cl& mu,
                             const T_scale_cl& sigma,
                             const T_skewness_cl& tau) {
  static constexpr const char* function
      = "skew_double_exponential_lcdf(OpenCL)";
  using T_partials_return
      = partials_return_t<T_y_cl, T_loc_cl, T_scale_cl, T_skewness_cl>;
  using std::isfinite;
  using std::isnan;
 
  check_consistent_sizes(function, "Random variable", y, "Location parameter",
                         mu, "Shape parameter", sigma, "Skewness parameter",
                         tau);
  const size_t N = max_size(y, mu, sigma, tau);
  if (N == 0) {
    return 1.0;
  }
 
  const auto& y_col = as_column_vector_or_scalar(y);
  const auto& mu_col = as_column_vector_or_scalar(mu);
  const auto& sigma_col = as_column_vector_or_scalar(sigma);
  const auto& tau_col = as_column_vector_or_scalar(tau);
 
  const auto& y_val = value_of(y_col);
  const auto& mu_val = value_of(mu_col);
  const auto& sigma_val = value_of(sigma_col);
  const auto& tau_val = value_of(tau_col);
 
  auto check_y_not_nan
      = check_cl(function, "Random variable", y_val, "not NaN");
  auto y_not_nan_expr = !isnan(y_val);
  auto check_mu_finite
      = check_cl(function, "Location parameter", mu_val, "finite");
  auto mu_finite_expr = isfinite(mu_val);
  auto check_sigma_positive_finite
      = check_cl(function, "Scale parameter", sigma_val, "positive finite");
  auto sigma_positive_finite_expr = 0.0 < sigma_val && isfinite(sigma_val);
  auto check_tau_bounded = check_cl(function, "Skewness parameter", tau_val,
                                    "in the interval [0, 1]");
  auto tau_bounded_expr = 0.0 < tau_val && tau_val <= 1.0;
 
  auto inv_sigma = elt_divide(1.0, sigma_val);
  auto y_m_mu = y_val - mu_val;
  auto diff_sign = sign(y_m_mu);
  auto diff_sign_smaller_0 = diff_sign < 0;
  auto abs_diff_y_mu = fabs(y_m_mu);
  auto abs_diff_y_mu_over_sigma = elt_multiply(abs_diff_y_mu, inv_sigma);
  auto expo
      = elt_multiply(diff_sign_smaller_0 + elt_multiply(diff_sign, tau_val),
                     abs_diff_y_mu_over_sigma);
  auto tau_minus_1 = tau_val - 1.0;
  auto inv_exp_2_expo_tau = elt_divide(1.0, exp(2.0 * expo) + tau_minus_1);
 
  auto lcdf_expr
      = colwise_sum(select(y_val <= mu_val, log(tau_val) - 2.0 * expo,
                           log1m_exp(log1m(tau_val) - 2.0 * expo)));
 
  auto cond = y_val < mu_val;
  auto y_deriv
      = select(cond, -2.0 * elt_multiply(inv_sigma, tau_minus_1),
               -2.0
                   * elt_multiply(elt_multiply(tau_minus_1, tau_val),
                                  elt_multiply(inv_sigma, inv_exp_2_expo_tau)));
  auto mu_deriv = -y_deriv;
  auto sigma_deriv = select(cond, 2.0 * elt_multiply(inv_sigma, expo),
                            -elt_divide(elt_multiply(y_deriv, expo), tau_val));
  auto tau_deriv = select(
      cond,
      elt_divide(1.0, tau_val)
          + 2.0 * elt_multiply(elt_multiply(inv_sigma, y_m_mu), diff_sign),
      elt_multiply(sigma_val - 2.0 * elt_multiply(tau_minus_1, y_m_mu),
                   elt_multiply(inv_sigma, inv_exp_2_expo_tau)));
 
  matrix_cl<double> lcdf_cl;
  matrix_cl<double> y_deriv_cl;
  matrix_cl<double> mu_deriv_cl;
  matrix_cl<double> sigma_deriv_cl;
  matrix_cl<double> tau_deriv_cl;
 
  results(check_y_not_nan, check_mu_finite, check_sigma_positive_finite,
          check_tau_bounded, lcdf_cl, y_deriv_cl, mu_deriv_cl, sigma_deriv_cl,
          tau_deriv_cl)
      = expressions(y_not_nan_expr, mu_finite_expr, sigma_positive_finite_expr,
                    tau_bounded_expr, lcdf_expr,
                    calc_if<is_autodiff_v<T_y_cl>>(y_deriv),
                    calc_if<is_autodiff_v<T_loc_cl>>(mu_deriv),
                    calc_if<is_autodiff_v<T_scale_cl>>(sigma_deriv),
                    calc_if<is_autodiff_v<T_skewness_cl>>(tau_deriv));
 
  T_partials_return lcdf = from_matrix_cl(lcdf_cl).sum();
 
  auto ops_partials
      = make_partials_propagator(y_col, mu_col, sigma_col, tau_col);
 
  if constexpr (is_autodiff_v<T_y_cl>) {
    partials<0>(ops_partials) = std::move(y_deriv_cl);
  }
  if constexpr (is_autodiff_v<T_loc_cl>) {
    partials<1>(ops_partials) = std::move(mu_deriv_cl);
  }
  if constexpr (is_autodiff_v<T_scale_cl>) {
    partials<2>(ops_partials) = std::move(sigma_deriv_cl);
  }
  if constexpr (is_autodiff_v<T_skewness_cl>) {
    partials<3>(ops_partials) = std::move(tau_deriv_cl);
  }
 
  return ops_partials.build(lcdf);
}
 
}  // namespace math
}  // namespace stan
#endif
#endif