Skip to contents

Calculate information dynamics using PPIDyOM

ppidyom.Rd

This function calls ppidyom on arbitrary vectors of data, or humdrumR data.

Usage

ppidyom(...)

Arguments

...

One or more input vectors, all the same length.

maxN

Maximum N-gram length to compute. Defaults to 5.

alphabet

The set of possible input values. By default, cartesian product of input vectors.

model_type

Which memory component(s) to use:

  • "stm" — short-term memory, within each shortTermGroups group only.

  • "ltm" — long-term memory, trained across longTermGroups.

  • "both" — STM + LTM blended.

  • "ltm+"/"both+" — as "ltm"/"both", but LTM updates online group by group.

ppm_type

PPM estimation method:

  • "interpolation" — weighted sum across all n-gram orders.

  • "backoff" — falls through orders from longest to shortest matching context.

shortTermArgs

List of STM settings:

  • lambda — escape method, one of "A"/"B"/"C"/"D"/"X" (default "C"); see the Escape method section of the Parameter Correspondence vignette.

  • exclusion — logical; exclude symbols already assigned a probability at a higher order (default TRUE).

  • update_exclusion — logical; stop updating lower-order counts once a higher order already matched at this timestep (default TRUE).

longTermArgs

List of LTM settings: same as shortTermArgs, plus start_token (whether to count beginning-of-sequence positions).

longTermGroups

Groups for long term training (usually pieces).

shortTermGroups

Groups for short term (local) application (usually parts within a piece).

b

Bias exponent for entropy-weighted blending, used only when model_type is "both"/"both+"; higher values favor whichever of STM/LTM is currently more confident.

idyom_base

Logical; use IDyOM's order-(-1) base distribution instead of the default shrinking-denominator base. See the Implementation Discrepancy vignette.

Examples

if (FALSE) { # \dontrun{
x <- c("A", "B", "A", "C", "A", "B", "A", "C", "A")
ppidyom(x, maxN = 3, model_type = "stm")
} # }