pkg/R/constructionModelesLassoRank.R

   1 #' constructionModelesLassoRank
   2 #'
   3 #' Construct a collection of models with the Lasso-Rank procedure.
   4 #'
   5 #' @param S output of selectVariables.R
   6 #' @param k number of components
   7 #' @param mini integer, minimum number of iterations in the EM algorithm, by default = 10
   8 #' @param maxi integer, maximum number of iterations in the EM algorithm, by default = 100
   9 #' @param X matrix of covariates (of size n*p)
  10 #' @param Y matrix of responses (of size n*m)
  11 #' @param eps real, threshold to say the EM algorithm converges, by default = 1e-4
  12 #' @param rank.min integer, minimum rank in the low rank procedure, by default = 1
  13 #' @param rank.max integer, maximum rank in the low rank procedure, by default = 5
  14 #' @param ncores Number of cores, by default = 3
  15 #' @param fast TRUE to use compiled C code, FALSE for R code only
  16 #' @param verbose TRUE to show some execution traces
  17 #'
  18 #' @return a list with several models, defined by phi, rho, pi, llh
  19 #'
  20 #' @export
  21 constructionModelesLassoRank <- function(S, k, mini, maxi, X, Y, eps, rank.min, rank.max,
  22   ncores, fast = TRUE, verbose = FALSE)
  23 {
  24   n <- dim(X)[1]
  25   p <- dim(X)[2]
  26   m <- dim(Y)[2]
  27   L <- length(S)
  28
  29   # Possible interesting ranks
  30   deltaRank <- rank.max - rank.min + 1
  31   Size <- deltaRank^k
  32   RankLambda <- matrix(0, nrow = Size * L, ncol = k + 1)
  33   for (r in 1:k)
  34   {
  35     # On veut le tableau de toutes les combinaisons de rangs possibles, et des
  36     # lambdas Dans la première colonne : on répète (rank.max-rank.min)^(k-1) chaque
  37     # chiffre : ça remplit la colonne Dans la deuxieme : on répète
  38     # (rank.max-rank.min)^(k-2) chaque chiffre, et on fait ça (rank.max-rank.min)^2
  39     # fois ...  Dans la dernière, on répète chaque chiffre une fois, et on fait ça
  40     # (rank.min-rank.max)^(k-1) fois.
  41     RankLambda[, r] <- rep(rank.min + rep(0:(deltaRank - 1), deltaRank^(r - 1),
  42       each = deltaRank^(k - r)), each = L)
  43   }
  44   RankLambda[, k + 1] <- rep(1:L, times = Size)
  45
  46   if (ncores > 1)
  47   {
  48     cl <- parallel::makeCluster(ncores, outfile = "")
  49     parallel::clusterExport(cl, envir = environment(), varlist = c("A1", "Size",
  50       "Pi", "Rho", "mini", "maxi", "X", "Y", "eps", "Rank", "m", "phi", "ncores",
  51       "verbose"))
  52   }
  53
  54   computeAtLambda <- function(index)
  55   {
  56     lambdaIndex <- RankLambda[index, k + 1]
  57     rankIndex <- RankLambda[index, 1:k]
  58     if (ncores > 1)
  59       require("valse")  #workers start with an empty environment
  60
  61     # 'relevant' will be the set of relevant columns
  62     selected <- S[[lambdaIndex]]$selected
  63     relevant <- c()
  64     for (j in 1:p)
  65     {
  66       if (length(selected[[j]]) > 0)
  67       {
  68         relevant <- c(relevant, j)
  69       }
  70     }
  71     if (max(rankIndex) < length(relevant))
  72     {
  73       phi <- array(0, dim = c(p, m, k))
  74       if (length(relevant) > 0)
  75       {
  76         res <- EMGrank(S[[lambdaIndex]]$Pi, S[[lambdaIndex]]$Rho, mini, maxi,
  77           X[, relevant], Y, eps, rankIndex, fast)
  78         llh <- c(res$LLF, sum(rankIndex * (length(relevant) - rankIndex +
  79           m)))
  80         phi[relevant, , ] <- res$phi
  81       }
  82       list(llh = llh, phi = phi, pi = S[[lambdaIndex]]$Pi, rho = S[[lambdaIndex]]$Rho)
  83
  84     }
  85   }
  86
  87   # For each lambda in the grid we compute the estimators
  88   out <- if (ncores > 1) {
  89     parLapply(cl, seq_len(length(S) * Size), computeAtLambda) } else {
  90         lapply(seq_len(length(S) * Size), computeAtLambda)
  91
  92   if (ncores > 1)
  93     parallel::stopCluster(cl)
  94
  95   out
  96 }