LOG-TYPES:  Types of log file records used by various programs.

Users do not usually need to know this information; it is mainly for
program maintainers.  However, these types are visible in the output
of the log-records program (see log-records.doc).
 

CLASS OF PROGRAMS          TYPE   DESCRIPTION


* Any                        r    State of random number generator


* Markov Chain Monte Carlo   i    Description of state/events at one iteration
                             o    List of Monte Carlo operations to apply
                             s    Explicitly set stepsizes (only some programs)
                             l    Value and how changed for slevel (optional)

  Dynamic MCMC               p    Values of "momentum" variables
                             t    Specification of how to compute trajectories

  Tempered MCMC              m    Schedule of temperatures and maybe biases
                             b    Current temperature and associated state

  Thermostated dynamics      h    Value of thermostat and corresponding momentum


* Specified distribution     d    Specification of distribution
                             q    Values of variables ("position" variables)


* Bivariate Gaussian         B    Specification of bivariate Gaussian 
                             X    Point from bivariate Gaussian distribution


* Multivariate Gaussian      M    Specification of multivariate Gaussian 
                             X    Point from multivariate Gaussian distribution


* Molecular dynamics         M    Specifications of molecular dynamics system
                             Q    Position coordinates of molecules


* Ising system               I    Specification of Ising system
                             S    State of spins in system
                             D    Direction to push 


* Statistical modeling       D    Data specifications for training and test sets
                             M    Model specification
                             V    Characteristics of model for survival data

  Neural network             A    Network architecture
                             F    Flags modifying architecture
                             P    Specification of priors for network
                             S    Hyperparameters ("sigmas")
                             W    Parameters of network ("weights")
                             Q    Parameters of quadratic approximation

  Gaussian process           P    Specification and priors
                             S    Hyperparameters
                             F    Case-by-case latent values
                             N    Case-by-case noise variances

  Mixture model              P    Specification and priors
                             S    Hyperparameters
                             I    Component indicators for training cases
                             O    Offset parameters for components
                             N    Noise variance parameters

  Diffusion tree model       P    Specification and priors
                             S    Hyperparameters and tree parameters
                             T    Tree divergence times
                             R    Parents of nodes in trees
                             L    Latent vectors for training cases
                             N    Locations of nodes in trees

  Source model               S    Source specification
                             T    Detector specification
                             F    Flow specification

  NMR model                  P    Model and prior specification
                             D    Parameters and latent values

        Copyright (c) 1995-2007 by Radford M. Neal