Built-in Inference Algorithms
Omega comes with a number of built in inference algorithms. You can of course develop your own.
Choosing a Sampling Algorithm
The appropriate sampling algorithm depends on the kind of model.
- If your model is not conditioned, or the conditions are not very restricting, use
- If your model is finite dimensional, continuous and unimodal use
- If your model is finite dimensional, continuous and multimodal use
- If your model is a mixture of discrete and continuous, or not of finite dimension, use
Conditional sampling is done with
rand and the algorithm are selected
Single Site Metropolis Hastings
Missing docstring for
NUTS. Check Documenter's build log for details.
Flux based Hamiltonian Monte Carlo Sampling