R/fun_get_component_depth.R

In daviddoret/haricot: Algorithm Composition, Manipulation And Analysis

require(checkmate);
#' get_component_depth
#'
#' Returns the depth of a component in a composite algorithm. \cr
#' \cr
#' \strong{Definition} \cr
#' The depth of a component is equal to: \cr
#' \itemize{
#' \item{1 if the component is a constant (ie it has no predecessors),}
#' \item{2 if the component is only directly linked to its parent composite,}
#' \item{the maximal depth of its predecessors + 1 otherwise.}
#' } \cr
#' \strong{Formula} \cr
#' \deqn{
#' depth(x)=\begin{cases}
#' 1, & \text{if}\ |N^{-}(x)| = 0 \\
#' 2, & \text{if}\ \forall y \in N^{-}(x), y = composite(x) \\
#' max(depth(N^{-}(x)))+1, & \text{otherwise}
#' \end{cases}
#' }
#' \strong{Note} \cr
#' In the haricot package, when we plot a composite algorithm,
#' we plot the input bits and output bits as vertices.
#' This representation is interesting because it shows the bit-level
#' interfaces between components.
#' But the depth concept does not consider input bits and output bits,
#' and only consider components as single units (the algorithm and its surrounding bits).
#' The reason for this is to account for the algorithmic complexity or size, rather than the detailed bit mechanics.
#'
#' @examples ...
#'
#' @param composite The composite within which we want to measure the depth of a component (R6 class algo_composite).
#' @param component The component algorithm that must be part of the composite and whose depth we would like to know (R6 class algo_base | character vector of the component ID).
#' @return The depth of the component (integer).
#' @export
get_component_depth = function(composite, component, ...) {

  checkmate::assert_r6(composite, "algo_composite");
  checkmate::assert_r6(component, "algo_base");

  if(component$get_dim_i() == 0){
    return(1);
  } else if(composite$get_algo_id() == component$get_algo_id()){
    return(1);
  } else {
    p_list <- predecessors(composite, component, ...);
    depth <- NULL;
    for(p in p_list){
      depth <- max(get_component_depth(composite, p, ...) + 1, depth);
    };
    return(depth);
  };
};