#include "probabilitydistribution.h"
#include <string>

Include dependency graph for configdistributionhelper.h:

Functions
ProbabilityDistribution *	getDistributionFromConfig (ConfigSettings &config, GslRandomNumberGenerator *pRndGen, const std::string &prefix)
	This is a helper function to more easily specify a particular 1D probability distribution in a config file. More...

Function Documentation

ProbabilityDistribution* getDistributionFromConfig	(	ConfigSettings &	config,
		GslRandomNumberGenerator *	pRndGen,
		const std::string &	prefix
	)

This is a helper function to more easily specify a particular 1D probability distribution in a config file.

The config parameter specifies the settings read from a config file, the pRndGen parameter is the random number generator to base the random numbers on and prefix specifies the start of the key in the config file.

For example, if the prefix is 'test', this function will first look for a key called 'test.dist.type'. Depending on its value, other keys for the parameters of the probability distribution will be sought. A newly created instance of the probability distribution will be returned on success, the program will abort when there's an error.

An overview of the currently supported distributions and their parameters (values are just examples):

Not really a distribution, but something you can use to set a specific value:

test.dist.type = fixed
test.dist.fixed.value = 5.0

A uniform distribution:

test.dist.type = uniform
test.dist.uniform.min = 4.0
test.dist.uniform.max = 5.0

Beta distribution,

$\textrm{prob}(x) = \frac{a+b}{\Gamma(a)\Gamma(b)} \left(\frac{x-min}{max-min}\right)^{a-1} \left( 1 - \frac{x-min}{max-min} \right)^{b-1} \frac{1}{max-min}$

test.dist.type = beta
test.dist.beta.a = 2
test.dist.beta.b = 2
test.dist.beta.min = -2
test.dist.beta.max = 3

Gamma distribution,

$\textrm{prob}(x) = \frac{ x^{a-1} \exp\left(-\frac{x}{b} \right) } {b^a \Gamma(a) }$

test.dist.type = gamma
test.dist.gamma.a = 1
test.dist.gamma.b = 1

Log-normal distribution,

$\textrm{prob}(x) = \frac{1}{x s \sqrt{2 \pi} } \exp\left(-\frac{(\textrm{ln}(x)-z)^2}{2 s^2}\right)$

test.dist.type = lognormal
test.dist.lognormal.zeta = 0
test.dist.lognormal.sigma = 0.5

Normal distribution,

$\textrm{prob}(x) = \frac{1}{\sigma \sqrt{2 \pi} } \exp\left(-\frac{(x-\mu)^2}{2 \sigma^2}\right)$

test.dist.type = normal
test.dist.normal.mu = 1
test.dist.normal.sigma = 2.5

Functions

Function Documentation