feng-li/dng

Distributions and Gradients

dng

The R package dng (distributions and gradients) provides efficient algorithms for density, distribution function, quantile function and random generation for the split normal and split-t distributions, and computes their mean, variance, skewness and kurtosis for the two distributions.

Installation

CRAN version

install.packages("dng")

Development version

You can install the development version of dng package from GitHub Repository with:

devtools::install_github("feng-li/dng")

Usage

Split normal distributions

library(dng)
n <- 3
mu <- c(0,1,2)
sigma <- c(1,2,3)
lmd <- c(1,2,3)
q0 <- rsplitn(n, mu, sigma, lmd)
d0 <- dsplitn(q0, mu, sigma, lmd, logarithm = FALSE)
p0 <- psplitn(q0, mu, sigma, lmd)
q1 <- qsplitn(p0,mu, sigma, lmd)
all.equal(q0, q1)

Split-t distributions

n <- 3
mu <- c(0,1,2)
sigma <- c(1,2,3)
lmd <- c(1,2,3)
q0 <- rsplitn(n, mu, sigma, lmd)
d0 <- dsplitn(q0, mu, sigma, lmd, logarithm = FALSE)
p0 <- psplitn(q0, mu, sigma, lmd)
q1 <- qsplitn(p0,mu, sigma, lmd)
all.equal(q0, q1)

References

• Li, Feng, Villani, Mattias, & Kohn, Robert. (2010). "Flexible modeling of conditional distributions using smooth mixtures of asymmetric student t densities". Journal of Statistical Planning & Inference, 140(12), 3638-3654.

License

This package is free and open source software, licensed under GPL-3.

Acknowledgements

Feng Li is supported by the National Natural Science Foundation of China.