The “stan” Function in R

• Package: rstan

• Purpose: Interface to Stan for Bayesian inference

• General class: Statistical modeling

• Required argument(s):

• model_code or file: Stan model code or file containing the Stan model.

• data: List or environment containing data for the Stan model.

• Notable optional arguments:

• pars: Parameters to monitor.

• chains: Number of Markov chains.

• cores: Number of cores to use for parallel computation.

• iter: Number of iterations.

• warmup: Number of warmup iterations.

• thin: Thinning parameter.

• seed: Random seed for reproducibility.

• algorithm: Sampling algorithm (e.g., “NUTS”).

• control: Control parameters for the sampling algorithm.

• Example:

• # Load the rstan library
  library(rstan)
  
  # Define the Stan model
  model_code <- '
  data {
  int<lower=0> N;
  real y[N];
  }
  parameters {
  real mu;
  real<lower=0> sigma;
  }
  model {
  y ~ normal(mu, sigma);
  }
  '
  
  # Generate synthetic data
  N <- 100
  data <- list(N = N, y = rnorm(N, 0, 1))
  
  # Run the Stan model
  fit <- stan(model_code = model_code, data = data, chains = 4, iter = 1000)
  summary(fit)

• This example demonstrates how to use the “stan” function from the rstan package to perform Bayesian inference using a simple normal distribution model. It involves defining the Stan model, preparing data, running the model with optional arguments like chains and iterations, and then summarizing the results.