Dakota Reference Manual  Version 6.4
Large-Scale Engineering Optimization and Uncertainty Analysis
 All Pages
chain_samples


Number of Markov Chain Monte Carlo posterior samples

Specification

Alias: samples

Argument(s): INTEGER

Default: method-dependent

Description

The chain_samples keyword indicates the number of draws from the posterior distribution to perform. When an emulator is active, this will be the number of samples on the constructed surrogate model.

Default Behavior

The default number of chain samples is method-dependent. QUESO methods use 48576. DREAM uses (number of generations) x (number of chains).

Usage Tips

MCMC methods typically require a large number of chain samples to converge, often thousands to millions.

Examples

method
  bayes_calibration queso
    chain_samples = 20000