struct gpr_model_config_t

Overview

GP/TP model configuration. More…

#include <gpr_optim_c.h>
 
struct gpr_model_config_t {
    // fields
 
    double sigma2;
    double jitter_sigma2;
    double magn_sigma2;
    double const_sigma2;
    double prior_mu;
    double prior_nu;
    double prior_s2;
    gpr_optim_alg_t optim_alg;
    int max_iter;
    double tol_func;
    double tol_sol;
    int report_level;
    double tp_hyperprior_a;
    double tp_hyperprior_b;
};

Detailed Documentation

GP/TP model configuration. All fields have sensible defaults via gpr_model_config_defaults().

Fields

double sigma2

Observation noise variance. Default: 1e-8

double jitter_sigma2

Diagonal jitter. Default: 1e-5

double magn_sigma2

Signal variance (0=auto). Default: 0

double const_sigma2

Constant covariance term. Default: 0

double prior_mu

Prior mean. Default: 0

double prior_nu

Prior degrees of freedom. Default: 20

double prior_s2

Prior scale. Default: 1

gpr_optim_alg_t optim_alg

Optimization algorithm. Default: SCG

int max_iter

Max optimizer iterations. Default: 400

double tol_func

Function tolerance. Default: 1e-4

double tol_sol

Solution tolerance. Default: 1e-4

int report_level

0=silent, 1=summary, 2=verbose. Default: 0

double tp_hyperprior_a

TP inverse-gamma shape. Default: 1e-20

double tp_hyperprior_b

TP inverse-gamma scale. Default: 1e-20