SGMCMC

Functions for running standard SGMCMC algorithms

sgld

Stochastic Gradient Langevin Dynamics

sghmc

Stochastic Gradient Hamiltonian Monte Carlo

sgnht

Stochastic Gradient Nose Hoover Thermostat

SGMCMC with Control Variates

Functions for running SGMCMC algorithms with improved efficiency using control variates

sgldcv

Stochastic Gradient Langevin Dynamics with Control Variates

sghmccv

Stochastic Gradient Hamiltonian Monte Carlo with Control Variates

sgnhtcv

Stochastic Gradient Nose Hoover Thermostat with Control Variates

Run SGMCMC algorithms step by step

Functions to run SGMCMC algorithms step by step within a user-defined loop, similar to standard TensorFlow optimization methods. Useful for chains with a high storage cost

sgldSetup

Create an sgld object

sgldcvSetup

Create an sgldcv object

sghmcSetup

Create an sghmc object

sghmccvSetup

Create an sghmccv object

sgnhtSetup

Create an sgnht object

sgnhtcvSetup

Create an sgnhtcv object

initSess

Initialise TensorFlow session and sgmcmc algorithm

sgmcmcStep

Single step of sgmcmc

getParams

Get current parameter values

Installation

Install python packages depended on by sgmcmc, TensorFlow and TensorFlow Probability.

installTF

Install TensorFlow and TensorFlow Probability

Datasets

Download and load datasets used in the examples and vignettes

getDataset

Load example datasets