1. The document discusses Bayesian model choice and compares multiple computational methods for approximating model evidence, including importance sampling, bridge sampling, and nested sampling.
2. It explains that Bayesian model choice involves probabilizing models and parameters, and computing the evidence for each model, which is proportional to the marginal likelihood.
3. One method covered is bridge sampling, which uses importance functions to approximate the Bayes factor for comparing two models. Optimal bridge sampling chooses an auxiliary function that minimizes the variance of the estimate.