Next: Writing SLPs
Up: New model spaces
Previous: Likelihood
Contents
This is not a necessary step for MCMCMS but it is used when
visualisation of the current and proposed models is requested.
You simply need to define
display_model( +Title, +Model, +Llhood ).
is the title of the model, i.e. where it is the current
or the proposal of iteration
.
is the representation of
the model to be displayed, and
is its log-likelihood.
When the predicate is entered all three arguments should be
instantiated appropriately, with the possible exception of
.
This depends on
whether model_llhood/2, see preceeding sub-Section,
instantiated the likelihood correctly. Your program should
be aware and take appropriate action if that is not the case
(e.g. not display the log-likelihood).
The main route for displaying models is via the run/1
option. This provides a hookable mechanism, the details of which
can be found in the sources: mcmcms/run.pl. Here we document
an implemented hook invoked by
display(next_display_at).
This always prints the first pair of models (current and proposed)
and interrogates the user for when the next display should be.
The number 1 and any non integer input will result
in the display of the next pair of models, while 0 will
halt the execution. A positive integer will
display the Nth comparison from the current position. A negative
integer will skip the Nth jump from the current position.
Next: Writing SLPs
Up: New model spaces
Previous: Likelihood
Contents
Nicos Angelopoulos
2008-06-02