neg_binomial_2_log_glm_lpmf() [1/3]

template<bool propto, typename T_y_cl , typename T_x_cl , typename T_alpha_cl , typename T_beta_cl , typename T_phi_cl , require_all_prim_or_rev_kernel_expression_t< T_x_cl, T_y_cl, T_alpha_cl, T_beta_cl, T_phi_cl > * = nullptr>

return_type_t< T_x_cl, T_alpha_cl, T_beta_cl, T_phi_cl > stan::math::neg_binomial_2_log_glm_lpmf ( const T_y_cl &  y, const T_x_cl &  x, const T_alpha_cl &  alpha, const T_beta_cl &  beta, const T_phi_cl &  phi  )

inline

Returns the log PMF of the Generalized Linear Model (GLM) with Negative-Binomial-2 distribution and log link function.

The idea is that neg_binomial_2_log_glm_lpmf(y, x, alpha, beta, phi) should compute a more efficient version of neg_binomial_2_log_lpmf(y, alpha + x * beta, phi) by using analytically simplified gradients. If containers are supplied, returns the log sum of the probabilities. This is an overload of the GLM in prim/prob/neg_binomial_2_log_glm_lpdf.hpp that is implemented in OpenCL.

Template Parameters

T_y_cl	type of independent variable; this can be a matrix_cl vector of intercepts or a single value (wich will be broadcast - used for all instances);
T_x_cl	type of the design matrix
T_alpha_cl	type of the intercept(s); this can be a vector (of the same length as y) of intercepts or a single value (for models with constant intercept);
T_beta_cl	type of the weight vector; this can also be a scalar;
T_phi_cl	type of the (positive) precision(s); this can be a vector (of the same length as y, for heteroskedasticity) or a scalar.

Parameters

y	failures count scalar or vector parameter on OpenCL device. If it is a scalar it will be broadcast - used for all instances.
x	design matrix on OpenCL device. This overload does not support broadcasting of a row vector x!
alpha	intercept (in log odds)
beta	weight vector
phi	(vector of) precision parameter(s)

Returns

log probability or log sum of probabilities

Exceptions

std::invalid_argument	if container sizes mismatch.
std::domain_error	if x, beta or alpha is infinite.
std::domain_error	if phi is infinite or non-positive.
std::domain_error	if y is negative.

Definition at line 69 of file neg_binomial_2_log_glm_lpmf.hpp.