DFT-GEN:  Generate parameters randomly from the prior, or fix them.

Dft-gen writes a series of independently-generated values for the
overall hyperparameters and diffusion tree parameters to a log file.


    dft-gen log-file [ max-index ] [ "fix" [ SD-value ] ]

Records of parameters with indexes from zero up to the indicated index
are generated (the default is max-index of zero).  If the log file
already contains records with some of these indexes, only records with
indexes greater than the last existing record are generated.

If just max-index is specified, the parameters are generated randomly
from the prior, using the random number seed taken from the log file
(eg, as specified by rand-seed).

The "fix" option is useful for initializing iterative programs.  If it
is given, the parameters are not generated at random, but are instead
set to fixed values.  The noise standard deviation hyperparameters and
the diffusion tree standard deviations will be fixed to the value
given after "fix", or at the location parameter of their prior
distribution, if no such value is specified.  The parameters of the
divergence function are fixed at the location parameter of their prior

No data model is required to use dft-gen, but if a model is specified,
the hyperparameters associated with it are generated as well.

Note that dft-gen does not generate any actual diffusion trees to go
with the training cases, nor does it generate latent values for
training cases.

            Copyright (c) 1995-2003 by Radford M. Neal