From 4cc632c9a1e1d93e9a43a402d1361f23afc50e5e Mon Sep 17 00:00:00 2001 From: Benjamin Auder Date: Mon, 3 Apr 2017 12:00:08 +0200 Subject: [PATCH] remove selectiontotale, parallelize main.R + add conditional verbose traces --- pkg/R/main.R | 94 ++++++++++++++++++++++++----------------- pkg/R/selectVariables.R | 6 ++- pkg/R/selectiontotale.R | 56 ------------------------ 3 files changed, 59 insertions(+), 97 deletions(-) delete mode 100644 pkg/R/selectiontotale.R diff --git a/pkg/R/main.R b/pkg/R/main.R index f080954..7b78a15 100644 --- a/pkg/R/main.R +++ b/pkg/R/main.R @@ -22,55 +22,59 @@ #' @export valse = function(X,Y,procedure = 'LassoMLE',selecMod = 'DDSE',gamma = 1,mini = 10, maxi = 50,eps = 1e-4,kmin = 2,kmax = 2, - rang.min = 1,rang.max = 10) + rang.min = 1,rang.max = 10, ncores_k=1, ncores_lambda=3, verbose=FALSE) { - #################### - # compute all models - #################### - p = dim(X)[2] m = dim(Y)[2] n = dim(X)[1] - - model = list() - tableauRecap = array(0, dim=c(1000,4)) - cpt = 0 - print("main loop: over all k and all lambda") - - for (k in kmin:kmax) + + tableauRecap = list() + if (verbose) + print("main loop: over all k and all lambda") + + if (ncores_k > 1) { - print(k) - print("Parameters initialization") + cl = parallel::makeCluster(ncores_k) + parallel::clusterExport( cl=cl, envir=environment(), varlist=c("X","Y","procedure", + "selecMod","gamma","mini","maxi","eps","kmin","kmax","rang.min","rang.max", + "ncores_k","ncores_lambda","verbose","p","m","k","tableauRecap") ) + } + + # Compute model with k components + computeModel <- function(k) + { + if (ncores_k > 1) + require("valse") #nodes start with an empty environment + + if (verbose) + print(paste("Parameters initialization for k =",k)) #smallEM initializes parameters by k-means and regression model in each component, #doing this 20 times, and keeping the values maximizing the likelihood after 10 #iterations of the EM algorithm. - init = initSmallEM(k, X, Y) - phiInit <- init$phiInit - rhoInit <- init$rhoInit - piInit <- init$piInit - gamInit <- init$gamInit - grid_lambda <- computeGridLambda(phiInit, rhoInit, piInit, gamInit, X, Y, gamma, mini, maxi, eps) - + P = initSmallEM(k, X, Y) + grid_lambda <- computeGridLambda(P$phiInit, P$rhoInit, P$piInit, P$gamInit, X, Y, + gamma, mini, maxi, eps) + + # TODO: 100 = magic number if (length(grid_lambda)>100) grid_lambda = grid_lambda[seq(1, length(grid_lambda), length.out = 100)] - print("Compute relevant parameters") + + if (verbose) + print("Compute relevant parameters") #select variables according to each regularization parameter #from the grid: A1 corresponding to selected variables, and #A2 corresponding to unselected variables. - - params = selectiontotale(phiInit,rhoInit,piInit,gamInit,mini,maxi,gamma,grid_lambda,X,Y,1e-8,eps) - #params2 = selectVariables(phiInit,rhoInit,piInit,gamInit,mini,maxi,gamma,grid_lambda[seq(1,length(grid_lambda), by=3)],X,Y,1e-8,eps) - ## etrange : params et params 2 sont différents ... - selected <- params$selected - Rho <- params$Rho - Pi <- params$Pi - + S = selectVariables(P$phiInit,P$rhoInit,P$piInit,P$gamInit,mini,maxi,gamma, + grid_lambda,X,Y,1e-8,eps,ncores_lambda) + if (procedure == 'LassoMLE') { - print('run the procedure Lasso-MLE') + if (verbose) + print('run the procedure Lasso-MLE') #compute parameter estimations, with the Maximum Likelihood #Estimator, restricted on selected variables. - model[[k]] = constructionModelesLassoMLE(phiInit, rhoInit,piInit,gamInit,mini,maxi,gamma,X,Y,thresh,eps,selected) + model = constructionModelesLassoMLE(phiInit, rhoInit, piInit, gamInit, mini, + maxi, gamma, X, Y, thresh, eps, S$selected) llh = matrix(ncol = 2) for (l in seq_along(model[[k]])) llh = rbind(llh, model[[k]][[l]]$llh) @@ -79,12 +83,13 @@ valse = function(X,Y,procedure = 'LassoMLE',selecMod = 'DDSE',gamma = 1,mini = 1 } else { - print('run the procedure Lasso-Rank') + if (verbose) + print('run the procedure Lasso-Rank') #compute parameter estimations, with the Low Rank #Estimator, restricted on selected variables. - model = constructionModelesLassoRank(Pi, Rho, mini, maxi, X, Y, eps, - A1, rank.min, rank.max) - + model = constructionModelesLassoRank(S$Pi, S$Rho, mini, maxi, X, Y, eps, A1, + rank.min, rank.max) + ################################################ ### Regarder la SUITE phi = runProcedure2()$phi @@ -98,13 +103,24 @@ valse = function(X,Y,procedure = 'LassoMLE',selecMod = 'DDSE',gamma = 1,mini = 1 Phi[, , 1:k,-(1:(dim(Phi2)[4]))] <<- phi } } - tableauRecap[(cpt+1):(cpt+length(model[[k]])), ] = matrix(c(LLH, D, rep(k, length(model[[k]])), 1:length(model[[k]])), ncol = 4) - cpt = cpt+length(model[[k]]) + tableauRecap[[k]] = matrix(c(LLH, D, rep(k, length(model[[k]])), 1:length(model[[k]])), ncol = 4)) } - print('Model selection') + + model <- + if (ncores_k > 1) + parLapply(cl, kmin:kmax, computeModel) + else + lapply(kmin:kmax, computeModel) + if (ncores_k > 1) + parallel::stopCluster(cl) + + if (verbose) + print('Model selection') + tableauRecap = do.call( rbind, tableaurecap ) #stack list cells into a matrix tableauRecap = tableauRecap[rowSums(tableauRecap[, 2:4])!=0,] tableauRecap = tableauRecap[(tableauRecap[,1])!=Inf,] data = cbind(1:dim(tableauRecap)[1], tableauRecap[,2], tableauRecap[,2], tableauRecap[,1]) + require(capushe) modSel = capushe(data, n) indModSel <- diff --git a/pkg/R/selectVariables.R b/pkg/R/selectVariables.R index b4fc0ab..869e7bf 100644 --- a/pkg/R/selectVariables.R +++ b/pkg/R/selectVariables.R @@ -14,6 +14,7 @@ #' @param Y matrix of responses #' @param thres threshold to consider a coefficient to be equal to 0 #' @param tau threshold to say that EM algorithm has converged +#' @param ncores Number or cores for parallel execution (1 to disable) #' #' @return a list of outputs, for each lambda in grid: selected,Rho,Pi #' @@ -22,7 +23,7 @@ #' @export #' selectVariables = function(phiInit,rhoInit,piInit,gamInit,mini,maxi,gamma,glambda, - X,Y,thresh,tau, ncores=1) #ncores==1 ==> no // + X,Y,thresh,tau, ncores=3) { if (ncores > 1) { @@ -54,7 +55,8 @@ selectVariables = function(phiInit,rhoInit,piInit,gamInit,mini,maxi,gamma,glambd out <- if (ncores > 1) parLapply(cl, glambda, computeCoefs) - else lapply(glambda, computeCoefs) + else + lapply(glambda, computeCoefs) if (ncores > 1) parallel::stopCluster(cl) diff --git a/pkg/R/selectiontotale.R b/pkg/R/selectiontotale.R deleted file mode 100644 index 2cdac38..0000000 --- a/pkg/R/selectiontotale.R +++ /dev/null @@ -1,56 +0,0 @@ -#Return a list of outputs, for each lambda in grid: selected,Rho,Pi -selectiontotale = function(phiInit,rhoInit,piInit,gamInit,mini,maxi,gamma,glambda,X,Y,thresh,tau, parallel = FALSE){ - if (parallel) { - require(parallel) - cl = parallel::makeCluster( parallel::detectCores() / 4) # <-- ça devrait être un argument - parallel::clusterExport(cl=cl, - varlist=c("phiInit","rhoInit","gamInit","mini","maxi","glambda","X","Y","thresh","tau"), - envir=environment()) - #Pour chaque lambda de la grille, on calcule les coefficients - out = parLapply(cl, 1:length(glambda), function(lambdaIndex) - { - params = - EMGLLF(phiInit,rhoInit,piInit,gamInit,mini,maxi,gamma,glambda[lambdaIndex],X,Y,tau) - - p = dim(phiInit)[1] - m = dim(phiInit)[2] - #selectedVariables: list where element j contains vector of selected variables in [1,m] - selectedVariables = lapply(1:p, function(j) { - #from boolean matrix mxk of selected variables obtain the corresponding boolean m-vector, - #and finally return the corresponding indices - seq_len(m)[ apply( abs(params$phi[j,,]) > thresh, 1, any ) ] - }) - - list("selected"=selectedVariables,"Rho"=params$Rho,"Pi"=params$Pi) - }) - parallel::stopCluster(cl) - } - else { - selectedVariables = list() - Rho = list() - Pi = list() - cpt = 1 - #Pour chaque lambda de la grille, on calcule les coefficients - for (lambdaIndex in 1:length(glambda)){ - print(lambdaIndex) - params = - EMGLLF(phiInit,rhoInit,piInit,gamInit,mini,maxi,gamma,glambda[lambdaIndex],X,Y,tau) - p = dim(phiInit)[1] - m = dim(phiInit)[2] - #selectedVariables: list where element j contains vector of selected variables in [1,m] - if (sum(params$phi) != 0){ - selectedVariables[[cpt]] = sapply(1:p, function(j) { - #from boolean matrix mxk of selected variables obtain the corresponding boolean m-vector, - #and finally return the corresponding indices - c(seq_len(m)[ apply( abs(params$phi[j,,]) > thresh, 1, any ) ], rep(0, m-length(seq_len(m)[ apply( abs(params$phi[j,,]) > thresh, 1, any ) ] ) )) - }) - if (length(unique(selectedVariables)) == length(selectedVariables)){ - Rho[[cpt]] = params$rho - Pi[[cpt]] = params$pi - cpt = cpt+1 - } - } - } - list("selected"=selectedVariables,"Rho"=Rho,"Pi"=Pi) - } -} \ No newline at end of file -- 2.44.0