Obsidian
Probabilistic Geophysical Joint Inversion
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Pages
stateline::MCMCSettings Struct Reference

Settings for Markov Chain Monte Carlo simulations. More...

#include <settings.hpp>

Public Attributes

uint chains
 The number of chains per stack.
 
uint stacks
 The number of stacks used in the simulation.
 
uint wallTime
 Maximum time the simulation should run for.
 
uint swapInterval
 The number of samples between swap attempts.
 
uint adaptionLength
 The number of samples from the beginning.
 
uint cacheLength
 The number of samples kept in memory. The MCMC uses a sliding buffer which flushes to disk old samples when the cache is full.
 
double initialSigmaFactor
 The initial proposal width factor. Initially, the proposal width of a chain will be initialSigmaFactor times the proposal width of the chain below (colder) than it.
 
double proposalInitialSigma
 The initial proposal width of the coldest chains.
 
double proposalMinFactor
 The minimum proposal adaption factor allowed. This controls how much the proposal width can change in one adaption step.
 
double proposalMaxFactor
 The maximum proposal adaption factor allowed. This controls how much the proposal width can change in one adaption step.
 
double proposalOptimalAccept
 The optimal acceptance rate for any of the chains.
 
double proposalAdaptRate
 The rate at which the proposal width changes when it adapts.
 
uint proposalAdaptInterval
 The number of samples between each attempt to adapt the proposal width.
 
double betaOptimalSwapRate
 The optimal swap rate for any of the chains.
 
double betaAdaptRate
 The rate at which the inverse temperature changes when it adapts.
 
double betaMinFactor
 The minimum beta adaption factor allowed. This controls how much the temperature can change in one adaption step.
 
double betaMaxFactor
 The minimum beta adaption factor allowed. This controls how much the temperature can change in one adaption step.
 
uint betaAdaptInterval
 The number of samples between each attempt to adapt the temperature.
 
double initialTempFactor
 The initial temperature ladder factor.
 

Detailed Description

Settings for Markov Chain Monte Carlo simulations.


The documentation for this struct was generated from the following file: