make.multimodal.dist: Create a distribution object for a random mixture of...

In SamplerCompare: A Framework for Comparing the Performance of MCMC Samplers

Create a distribution object for a random mixture of Gaussians

Description

Create a distribution object for a random mixture of Gaussians

Usage

make.multimodal.dist(nmodes, ndim, cube.size)

Arguments

nmodes	The number of components in the mixture model.
ndim	The dimension of the model.
cube.size	The edge length of the hypercube in which the modes are distributed.

Details

Defines a distribution object for a mixture of random Gaussians. The means of the the nmodes Gaussians are randomly distributed over an ndim-dimensional hypercube with one corner at the origin and the opposite cube.size away in each positive direction. The same random seed is temporarily set when drawing modes, so every time this function is called with the same parameters, the resulting distribution is the same.

This is included as a test case for comparing how MCMC methods perform on multimodal distributions.

Value

A scdist object. For convenience, the modes element is a matrix containing the modes as rows.

See Also

make.dist, compare.samplers

SamplerCompare documentation built on April 24, 2023, 9:09 a.m.