# rmv: Multivariate random number generation In rmv: Multivariate random number generation

## Description

Generates correlated random variables by the matrix square-root method.

## Usage

 `1` ```rmv(n, covmat, rfunc = rnorm, method = c("chol", "eigen"), ...) ```

## Arguments

 `n` Number of random multivariate draws. `covmat` The covariance matrix with which to obtain correlations. `rfunc` Function that returns a vector of random draws from a particular distribution. The first argument of `rfunc` must give the number of random draws to make. `method` The method for taking the matrix square-root. Currently only `'chol'` and `'eigen'` are allowed for the Cholesky and eigen decomposition methods. See `msr`. `...` Further arguments to pass to `rfunc`.

## Details

The number, `m`, of variables generated per each of the `n` observations is equal to the number of rows (or columns) of `covmat`. `rfunc` is used to fill an `n` by `m` matrix of uncorrelated observations. This matrix is multiplied by the matrix square-root (`msr`) to obtain a matrix of `m` correlated random variables.

## Value

A matrix with rows giving observations and columns giving correlated variables.

## Examples

 ```1 2 3 4 5``` ```set.seed(1) cm <- cor(matrix(rnorm(12), 6, 2)) x <- rmv(100, cm, rbinom, size = 100, prob = 0.7) pairs(x) cm ```

rmv documentation built on May 31, 2017, 5:05 a.m.