https://github.com/bisaloo/simplemh

Simple R implementation of the Metropolis-Hastings MCMC Algorithm
https://github.com/bisaloo/simplemh

Last synced: 11 days ago
JSON representation

Simple R implementation of the Metropolis-Hastings MCMC Algorithm

• Host: GitHub
• URL: https://github.com/bisaloo/simplemh
• Owner: Bisaloo
• Created: 2021-01-12T14:10:16.000Z (almost 4 years ago)
• Default Branch: main
• Last Pushed: 2024-02-09T09:15:06.000Z (11 months ago)
• Last Synced: 2024-11-24T09:11:46.509Z (29 days ago)
• Language: R
• Homepage: http://hugogruson.fr/simpleMH/
• Size: 1.5 MB
• Stars: 0
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.Rmd
  • Changelog: NEWS.md

Awesome Lists containing this project

README

        
---

output: github_document

---

```{r, include = FALSE}

knitr::opts_chunk$set(

  collapse = TRUE,

  comment = "#>",

  fig.path = "man/figures/README-",

  out.width = "100%"

)

```

# simpleMH

[![Lifecycle: maturing](https://img.shields.io/badge/lifecycle-maturing-blue.svg)](https://lifecycle.r-lib.org/articles/stages.html)

[![R-CMD-check](https://github.com/Bisaloo/simpleMH/actions/workflows/R-CMD-check.yaml/badge.svg)](https://github.com/Bisaloo/simpleMH/actions/workflows/R-CMD-check.yaml)

[![Codecov test coverage](https://codecov.io/gh/Bisaloo/simpleMH/branch/main/graph/badge.svg)](https://app.codecov.io/gh/Bisaloo/simpleMH?branch=main)

[![CRAN status](https://www.r-pkg.org/badges/version-ago/simpleMH)](https://CRAN.R-project.org/package=simpleMH)

This package offers a very bare-bones interface to use the Metropolis-Hastings

Monte Carlo Markov Chain algorithm. It is suitable for teaching and testing

purposes. For more advanced uses, you can check out the

[mcmcensemble](https://hugogruson.fr/mcmcensemble/) or

[adaptMCMC](https://github.com/scheidan/adaptMCMC) packages, which are designed

with a very similar interface, but often allow better convergence, especially

for badly scaled problems or highly correlated set of parameters.

## Installation

You can install this package from CRAN:

```{r, eval = FALSE}

install.packages("simpleMH")

```

or from my [r-universe](https://bisaloo.r-universe.dev/) (development version):

```{r, eval = FALSE}

install.packages("simpleMH", repos = "https://bisaloo.r-universe.dev")

```

## Example

```{r example}

library(simpleMH)

## a log-pdf to sample from

p.log <- function(x) {

  B <- 0.03                              # controls 'bananacity'

  -x[1]^2/200 - 1/2*(x[2]+B*x[1]^2-100*B)^2

}

res <- simpleMH(

  p.log,

  inits = c(0, 0),

  theta.cov = diag(2),

  max.iter = 5000,

  coda = TRUE # to be able to have nice plots and diagnostics with the coda pkg

)

```

Here is the resulting sampling landscape of `p.log()`:

```{r sampling-landscape, dpi = 150}

plot(as.data.frame(res$samples))

```

We can then use the [coda package](https://cran.r-project.org/package=coda) to

post-process the chain (burn-in, thinning, etc.), plot the trace and density,

or compute convergence diagnostics:

```{r coda-plots, dpi = 150}

plot(res$samples)

```