Handlers and Helpers

Utility functions and parameter handlers. More...

Collaboration diagram for Handlers and Helpers:

Subgroups
	Convolution Functions
	Functions for convolving time series.
	Delay Handlers
	Functions for delay distribution handling.
	PMF Handlers
	Functions for probability mass function handling.
	Parameter Handlers
	Functions for parameter management.

Functions
int	neg_binomial_2_safe_rng (real mu, real phi)

Description

Utility functions and parameter handlers.

This group contains utility functions and parameter handlers from convolve.stan, params.stan, delays.stan, and pmfs.stan.

Function Documentation

neg_binomial_2_safe_rng()

int neg_binomial_2_safe_rng	(	real	mu,
		real	phi )

#include </github/workspace/inst/stan/functions/observation_model.stan>

Custom safe version of the negative binomial sampler

This function generates random samples of the negative binomial distribution whilst avoiding numerical overflows. In particular:

• if the mu parameter is very small it always returns 0
• if the phi parameter is large it returns a sample from a Poisosn distribution
• if the gamma rate of the gamma-Poisson mixture used for simulating from the distribution is very large, it returns 1e8
• in all other cases it returns a sample from the negative binomial distribution

Parameters

mu	Real value for mean mu.
phi	Real value for phi.

Returns

A random sample

Definition at line 231 of file observation_model.stan.

                                               {
  if (mu < 1e-8) {
    return(0);
  } else if (phi > 1e4) {
    return(poisson_rng(mu > 1e8 ? 1e8 : mu));
  } else {
    real gamma_rate = gamma_rng(phi, phi / mu);
    return(poisson_rng(gamma_rate > 1e8 ? 1e8 : gamma_rate));
  }
}

Referenced by report_rng().

Here is the caller graph for this function: