ltm_options: LTM options
In pmcharrison/hvr: Harmonic Viewpoint Regression

Description Usage Arguments

Creates an option list for the long-term discrete viewpoint models in the viewpoint regression model. Most (all?) of these options are passed to the PPM model.

ltm_options(
  enabled = TRUE,
  order_bound = 10L,
  shortest_deterministic = TRUE,
  exclusion = TRUE,
  update_exclusion = FALSE,
  escape = "c"
)

`enabled`	(Logical scalar) Whether or not the model is enabled.
`order_bound`	(Integerish scalar) The model's Markov order bound. For example, an order bound of two means that the model makes predictions based on the two preceding symbols.
`shortest_deterministic`	(Logical scalar) If TRUE, the model will 'select' the shortest available order that provides a deterministic prediction, if such an order exists, otherwise defaulting to the longest available order. For a given prediction, if this rule results in a lower model order than would have otherwise been selected, then full counts (not update-excluded counts) will be used for the highest model order (but not for lower model orders). This behaviour matches the implementations of PPM* in Pearce (2005) and Bunton (1996).
`exclusion`	(Logical scalar) If TRUE, implements exclusion as defined in Pearce (2005) and Bunton (1996).
`update_exclusion`	(Logical scalar) If TRUE, implements update exclusion as defined in Pearce (2005) and Bunton (1996).
`escape`	(Character scalar) Takes values 'a', 'b', 'c', 'd', or 'ax', corresponding to the eponymous escape methods in Pearce (2005). Note that there is a mistake in the definition of escape method "AX" in Pearce (2005); the denominator of lambda needs to have 1 added. This is what we implement here. Note that Pearce's LISP implementation correctly adds 1 here, like us.