X-Git-Url: https://git.auder.net/?a=blobdiff_plain;f=epclust%2FR%2Fclustering.R;h=74d009e45495220304a0087554e619d3c9efc4e4;hb=492cd9e74a79cbcc0ecde55fa3071a44b7e463dc;hp=578b2f399035023eb68b0b004be10f18d87cac84;hpb=48108c3999d28d973443fa5e78f73a0a9f2bfc07;p=epclust.git diff --git a/epclust/R/clustering.R b/epclust/R/clustering.R index 578b2f3..74d009e 100644 --- a/epclust/R/clustering.R +++ b/epclust/R/clustering.R @@ -1,65 +1,169 @@ -# Cluster one full task (nb_curves / ntasks series) -clusteringTask = function(indices_clust) +#' @name clustering +#' @rdname clustering +#' @aliases clusteringTask1 computeClusters1 computeClusters2 +#' +#' @title Two-stage clustering, withing one task (see \code{claws()}) +#' +#' @description \code{clusteringTask1()} runs one full stage-1 task, which consists in +#' iterated stage 1 clustering (on nb_curves / ntasks energy contributions, computed +#' through discrete wavelets coefficients). \code{computeClusters1()} and +#' \code{computeClusters2()} correspond to the atomic clustering procedures respectively +#' for stage 1 and 2. The former applies the clustering algorithm (PAM) on a +#' contributions matrix, while the latter clusters a chunk of series inside one task +#' (~max nb_series_per_chunk) +#' +#' @param indices Range of series indices to cluster in parallel (initial data) +#' @param getContribs Function to retrieve contributions from initial series indices: +#' \code{getContribs(indices)} outpus a contributions matrix +#' @param contribs matrix of contributions (e.g. output of \code{curvesToContribs()}) +#' @inheritParams computeSynchrones +#' @inheritParams claws +#' +#' @return For \code{clusteringTask1()} and \code{computeClusters1()}, the indices of the +#' computed (K1) medoids. Indices are irrelevant for stage 2 clustering, thus +#' \code{computeClusters2()} outputs a matrix of medoids +#' (of size limited by nb_series_per_chunk) +NULL + +#' @rdname clustering +#' @export +clusteringTask1 = function( + indices, getContribs, K1, nb_series_per_chunk, ncores_clust=1, verbose=FALSE, parll=TRUE) { - cl_clust = parallel::makeCluster(ncores_clust) - parallel::clusterExport(cl_clust, - varlist=c("K1","K2","WER"), - envir=environment()) - repeat - { - nb_workers = max( 1, round( length(indices_clust) / nb_series_per_chunk ) ) - indices_workers = lapply(seq_len(nb_workers), function(i) { - upper_bound = ifelse( i 0) + if (parll) + { + cl = parallel::makeCluster(ncores_clust) + parallel::clusterExport(cl, varlist=c("getContribs","K1","verbose"), envir=environment()) + } + while (length(indices) > K1) { - curves = computeSynchrones(cl) - dists = computeWerDists(curves) - cl = computeClusters(dists, K2, diss=TRUE) + indices_workers = .spreadIndices(indices, nb_series_per_chunk) + if (parll) + indices = unlist( parallel::parLapply(cl, indices_workers, wrapComputeClusters1) ) + else + indices = unlist( lapply(indices_workers, wrapComputeClusters1) ) } - indices_chunk[cl] + if (parll) + parallel::stopCluster(cl) + + indices #medoids } -# Apply the clustering algorithm (PAM) on a coeffs or distances matrix -computeClusters = function(md, K, diss) +#' @rdname clustering +#' @export +computeClusters1 = function(contribs, K1) + cluster::pam(contribs, K1, diss=FALSE)$id.med + +#' @rdname clustering +#' @export +computeClusters2 = function(medoids, K2, + getRefSeries, nb_ref_curves, nb_series_per_chunk, ncores_clust=1,verbose=FALSE,parll=TRUE) { - if (!require(cluster, quietly=TRUE)) - stop("Unable to load cluster library") - cluster::pam(md, K, diss=diss)$id.med + synchrones = computeSynchrones(medoids, + getRefSeries, nb_ref_curves, nb_series_per_chunk, ncores_clust, verbose, parll) + distances = computeWerDists(synchrones, ncores_clust, verbose, parll) + medoids[ cluster::pam(distances, K2, diss=TRUE)$medoids , ] } -# Compute the synchrones curves (sum of clusters elements) from a clustering result -computeSynchrones = function(indices) +#' computeSynchrones +#' +#' Compute the synchrones curves (sum of clusters elements) from a matrix of medoids, +#' using L2 distances. +#' +#' @param medoids Matrix of medoids (curves of same legnth as initial series) +#' @param getRefSeries Function to retrieve initial series (e.g. in stage 2 after series +#' have been replaced by stage-1 medoids) +#' @param nb_ref_curves How many reference series? (This number is known at this stage) +#' @inheritParams claws +#' +#' @export +computeSynchrones = function(medoids, getRefSeries, + nb_ref_curves, nb_series_per_chunk, ncores_clust=1,verbose=FALSE,parll=TRUE) { - colSums( getData(indices) ) + computeSynchronesChunk = function(indices) + { + ref_series = getRefSeries(indices) + #get medoids indices for this chunk of series + for (i in seq_len(nrow(ref_series))) + { + j = which.min( rowSums( sweep(medoids, 2, ref_series[i,], '-')^2 ) ) + if (parll) + synchronicity::lock(m) + synchrones[j,] = synchrones[j,] + ref_series[i,] + counts[j,1] = counts[j,1] + 1 + if (parll) + synchronicity::unlock(m) + } + } + + K = nrow(medoids) + # Use bigmemory (shared==TRUE by default) + synchronicity to fill synchrones in // + synchrones = bigmemory::big.matrix(nrow=K,ncol=ncol(medoids),type="double",init=0.) + counts = bigmemory::big.matrix(nrow=K,ncol=1,type="double",init=0) + # Fork (// run) only on Linux & MacOS; on Windows: run sequentially + parll = (requireNamespace("synchronicity",quietly=TRUE) + && parll && Sys.info()['sysname'] != "Windows") + if (parll) + m <- synchronicity::boost.mutex() + + indices_workers = .spreadIndices(seq_len(nb_ref_curves), nb_series_per_chunk) + for (inds in indices_workers) + { + if (verbose) + { + cat(paste("--- Compute synchrones for indices range ", + min(inds)," -> ",max(inds),"\n", sep="")) + } + if (parll) + ignored <- parallel::mcparallel(computeSynchronesChunk(inds)) + else + computeSynchronesChunk(inds) + } + if (parll) + parallel::mccollect() + + mat_syncs = matrix(nrow=K, ncol=ncol(medoids)) + vec_count = rep(NA, K) + #TODO: can we avoid this loop? + for (i in seq_len(K)) + { + mat_syncs[i,] = synchrones[i,] + vec_count[i] = counts[i,1] + } + #NOTE: odds for some clusters to be empty? (when series already come from stage 2) + # ...maybe; but let's hope resulting K1' be still quite bigger than K2 + mat_syncs = sweep(mat_syncs, 1, vec_count, '/') + mat_syncs[ sapply(seq_len(K), function(i) all(!is.nan(mat_syncs[i,]))) , ] } -# Compute the WER distance between the synchrones curves -computeWerDist = function(curves) +#' computeWerDists +#' +#' Compute the WER distances between the synchrones curves (in rows), which are +#' returned (e.g.) by \code{computeSynchrones()} +#' +#' @param synchrones A matrix of synchrones, in rows. The series have same length as the +#' series in the initial dataset +#' @inheritParams claws +#' +#' @export +computeWerDists = function(synchrones, ncores_clust=1,verbose=FALSE,parll=TRUE) { - if (!require("Rwave", quietly=TRUE)) - stop("Unable to load Rwave library") - n <- nrow(curves) - delta <- ncol(curves) + n <- nrow(synchrones) + delta <- ncol(synchrones) #TODO: automatic tune of all these parameters ? (for other users) nvoice <- 4 - # noctave = 2^13 = 8192 half hours ~ 180 days ; ~log2(ncol(curves)) + # noctave = 2^13 = 8192 half hours ~ 180 days ; ~log2(ncol(synchrones)) noctave = 13 # 4 here represent 2^5 = 32 half-hours ~ 1 day #NOTE: default scalevector == 2^(0:(noctave * nvoice) / nvoice) * s0 (?) @@ -71,33 +175,107 @@ computeWerDist = function(curves) s0log = as.integer( (log2( s0*w0/(2*pi) ) - 1) * nvoice + 1.5 ) totnoct = noctave + as.integer(s0log/nvoice) + 1 - # (normalized) observations node with CWT - Xcwt4 <- lapply(seq_len(n), function(i) { - ts <- scale(ts(curves[i,]), center=TRUE, scale=scaled) + computeCWT = function(i) + { + if (verbose) + cat(paste("+++ Compute Rwave::cwt() on serie ",i,"\n", sep="")) + ts <- scale(ts(synchrones[i,]), center=TRUE, scale=scaled) totts.cwt = Rwave::cwt(ts,totnoct,nvoice,w0,plot=0) ts.cwt = totts.cwt[,s0log:(s0log+noctave*nvoice)] #Normalization sqs <- sqrt(2^(0:(noctave*nvoice)/nvoice)*s0) sqres <- sweep(ts.cwt,MARGIN=2,sqs,'*') sqres / max(Mod(sqres)) - }) + } - Xwer_dist <- matrix(0., n, n) + if (parll) + { + cl = parallel::makeCluster(ncores_clust) + parallel::clusterExport(cl, varlist=c("getContribs","K1","verbose"), envir=environment()) + } + + # (normalized) observations node with CWT + Xcwt4 <- + if (parll) + parallel::parLapply(cl, seq_len(n), computeCWT) + else + lapply(seq_len(n), computeCWT) + + if (parll) + parallel::stopCluster(cl) + + Xwer_dist <- bigmemory::big.matrix(nrow=n, ncol=n, type="double") fcoefs = rep(1/3, 3) #moving average on 3 values (TODO: very slow! correct?!) - for (i in 1:(n-1)) + if (verbose) + cat("*** Compute WER distances from CWT\n") + + computeDistancesLineI = function(i) { + if (verbose) + cat(paste(" Line ",i,"\n", sep="")) for (j in (i+1):n) { - #TODO: later, compute CWT here (because not enough storage space for 32M series) - # 'circular=TRUE' is wrong, should just take values on the sides; to rewrite in C + #TODO: 'circular=TRUE' is wrong, should just take values on the sides; to rewrite in C num <- filter(Mod(Xcwt4[[i]] * Conj(Xcwt4[[j]])), fcoefs, circular=TRUE) WX <- filter(Mod(Xcwt4[[i]] * Conj(Xcwt4[[i]])), fcoefs, circular=TRUE) WY <- filter(Mod(Xcwt4[[j]] * Conj(Xcwt4[[j]])), fcoefs, circular=TRUE) wer2 <- sum(colSums(num)^2) / sum( sum(colSums(WX) * colSums(WY)) ) + if (parll) + synchronicity::lock(m) Xwer_dist[i,j] <- sqrt(delta * ncol(Xcwt4[[1]]) * (1 - wer2)) Xwer_dist[j,i] <- Xwer_dist[i,j] + if (parll) + synchronicity::unlock(m) + } + Xwer_dist[i,i] = 0. + } + + parll = (requireNamespace("synchronicity",quietly=TRUE) + && parll && Sys.info()['sysname'] != "Windows") + if (parll) + m <- synchronicity::boost.mutex() + + for (i in 1:(n-1)) + { + if (parll) + ignored <- parallel::mcparallel(computeDistancesLineI(i)) + else + computeDistancesLineI(i) + } + Xwer_dist[n,n] = 0. + + if (parll) + parallel::mccollect() + + mat_dists = matrix(nrow=n, ncol=n) + #TODO: avoid this loop? + for (i in 1:n) + mat_dists[i,] = Xwer_dist[i,] + mat_dists +} + +# Helper function to divide indices into balanced sets +.spreadIndices = function(indices, nb_per_chunk) +{ + L = length(indices) + nb_workers = floor( L / nb_per_chunk ) + if (nb_workers == 0) + { + # L < nb_series_per_chunk, simple case + indices_workers = list(indices) + } + else + { + indices_workers = lapply( seq_len(nb_workers), function(i) + indices[(nb_per_chunk*(i-1)+1):(nb_per_chunk*i)] ) + # Spread the remaining load among the workers + rem = L %% nb_per_chunk + while (rem > 0) + { + index = rem%%nb_workers + 1 + indices_workers[[index]] = c(indices_workers[[index]], indices[L-rem+1]) + rem = rem - 1 } } - diag(Xwer_dist) <- numeric(n) - Xwer_dist + indices_workers }