#' @examples
#' #See example in ?plotHist
#' @export
-plotCoefs <- function(mr, params, idx)
+plotCoefs <- function(mr, params, idx, xtitle="Parameter")
{
L <- nrow(mr[[1]][[1]])
K <- ncol(mr[[1]][[1]])
avg_param <- as.double(params_hat)
std_param <- as.double(stdev)
matplot(cbind(params[o],avg_param[o],avg_param[o]+std_param[o],avg_param[o]-std_param[o]),
- col=1, lty=c(1,5,3,3), type="l", lwd=2, xlab="Parameter", ylab="Value")
+ col=1, lty=c(1,5,3,3), type="l", lwd=2, xlab=xtitle, ylab="")
#print(o) #not returning o to avoid weird Jupyter issue... (TODO:)
}
--- /dev/null
+optimBeta <- function(N, n, K, p, beta, b, link, ncores)
+{
+ library(morpheus)
+ res <- multiRun(
+ list(n=n,p=p,beta=beta,b=b,optargs=list(K=K,link=link)),
+ list(
+ # morpheus
+ function(fargs) {
+ library(morpheus)
+ K <- fargs$optargs$K
+ M <- computeMoments(fargs$X, fargs$Y)
+ fargs$optargs$M <- M
+ mu <- computeMu(fargs$X, fargs$Y, fargs$optargs)
+ op <- optimParams(K,link,fargs$optargs)
+ x_init <- c( rep(1/K,K-1), as.double(mu), rep(0,K) )
+ do.call(rbind, op$run(x_init))
+ },
+ # flexmix
+ function(fargs) {
+ library(flexmix)
+ source("../patch_Bettina/FLXMRglm.R")
+ K <- fargs$optargs$K
+ dat <- as.data.frame( cbind(fargs$Y,fargs$X) )
+ fm <- flexmix( cbind(V1, 1-V1) ~ .-V1, data=dat, k=K,
+ model = FLXMRglm(family = binomial(link = link)) )
+ p <- mean(fm@posterior[["scaled"]][,1])
+ out <- refit(fm)
+ beta_b <- sapply( seq_len(K), function(i) {
+ as.double( out@components[[1]][[i]][,1] )
+ } )
+ rbind(p, beta_b[2:nrow(beta_b),], beta_b[1,])
+ } ),
+ prepareArgs = function(fargs, index) {
+ library(morpheus)
+ io = generateSampleIO(fargs$n, fargs$p, fargs$beta, fargs$b, fargs$optargs$link)
+ fargs$X = io$X
+ fargs$Y = io$Y
+ fargs$optargs$K = ncol(fargs$beta)
+ fargs$optargs$M = computeMoments(io$X,io$Y)
+ fargs
+ }, N=N, ncores=ncores, verbose=TRUE)
+ p <- c(p, 1-sum(p))
+ for (i in 1:2)
+ res[[i]] <- alignMatrices(res[[i]], ref=rbind(p,beta,b), ls_mode="exact")
+ res
+}
+
+#model = binomial; default values:
+link = "logit"
+N <- 10
+d <- 2
+n <- 1e4
+ncores <- 1
+
+cmd_args <- commandArgs()
+for (arg in cmd_args)
+{
+ if (substr(arg,1,1)!='-') {
+ spl <- strsplit(arg,'=')[[1]]
+ if (spl[1] == "nc") {
+ ncores <- as.integer(spl[2])
+ } else if (spl[1] == "N") {
+ N <- as.integer(spl[2])
+ } else if (spl[1] == "n") {
+ n <- as.integer(spl[2])
+ } else if (spl[1] == "d") {
+ d <- as.integer(spl[2])
+ } else if (spl[1] == "link") {
+ link <- spl[2]
+ }
+ }
+}
+
+if (d == 2) {
+ K <- 2
+ p <- .5
+ b <- c(-.2, .5)
+ beta <- matrix( c(1,-2, 3,1), ncol=K )
+} else if (d == 5) {
+ K <- 2
+ p <- .5
+ b <- c(-.2, .5)
+ beta <- matrix( c(1,2,-1,0,3, 2,-3,0,1,0), ncol=K )
+} else if (d == 10) {
+ K <- 3
+ p <- c(.3, .3)
+ b <- c(-.2, 0, .5)
+ beta <- matrix( c(1,2,-1,0,3,4,-1,-3,0,2, 2,-3,0,1,0,-1,-4,3,2,0, -1,1,3,-1,0,0,2,0,1,-2), ncol=K )
+} else if (d == 20) {
+ K <- 3
+ p <- c(.3, .3)
+ b <- c(-.2, 0, .5)
+ beta <- matrix( c(1,2,-1,0,3,4,-1,-3,0,2,2,-3,0,1,0,-1,-4,3,2,0, -1,1,3,-1,0,0,2,0,1,-2,1,2,-1,0,3,4,-1,-3,0,2, 2,-3,0,1,0,-1,-4,3,2,0,1,1,2,2,-2,-2,3,1,0,0), ncol=K )
+}
+
+mr <- optimBeta(N, n, K, p, beta, b, link, ncores)
+mr_params <- list("N"=N, "n"=n, "K"=K, "d"=d, "link"=link,
+ "p"=c(p,1-sum(p)), "beta"=beta, "b"=b)
+
+save("mr", "mr_params", file=paste("multirun_",d,"_",link,".RData",sep=""))
--- /dev/null
+library(morpheus)
+
+#model = binomial
+K <- 2
+p <- .5
+b <- c(-.2, .5)
+# Default values:
+link = "logit"
+N <- 100
+d <- 2
+n <- 1e4
+ncores <- 1
+nstart <- 3 #nstart-1 random starting points for each MC run
+
+cmd_args <- commandArgs()
+for (arg in cmd_args)
+{
+ if (substr(arg,1,1)!='-')
+ {
+ spl <- strsplit(arg,'=')[[1]]
+ if (spl[1] == "nc") {
+ ncores <- as.integer(spl[2])
+ } else if (spl[1] == "N") {
+ N <- as.integer(spl[2])
+ } else if (spl[1] == "n") {
+ n <- as.integer(spl[2])
+ } else if (spl[1] == "d") {
+ d <- as.integer(spl[2])
+ } else if (spl[1] == "link") {
+ link <- spl[2]
+ } else if (spl[1] == "nstart") {
+ nstart <- spl[2]
+ }
+ }
+}
+betas <- list(
+ matrix( c(1,-2, 3,1), ncol=K ), #d=2
+ matrix( c(1,2,-1,0,3, 2,-3,0,1,0), ncol=K ), #d=5
+ matrix( c(1,2,-1,0,3,4,-1,-3,0,2, 2,-3,0,1,0,-1,-4,3,2,0), ncol=K ) ) #d=10
+beta <- betas[[ ifelse( d==2, 1, ifelse(d==5,2,3) ) ]]
+
+ms <- multiRun(
+ list(n=n,p=p,beta=beta,b=b,optargs=list(K=K,link=link,nstart=nstart)), list(
+ function(fargs) {
+ # 1 start
+ library(morpheus)
+ K <- fargs$optargs$K
+ op <- optimParams(K, fargs$optargs$link, fargs$optargs)
+ x_init <- c(rep(1/K,K-1), as.double(fargs$mu), rep(0,K))
+ do.call(rbind,op$run(x_init))
+ },
+ function(fargs) {
+ # B starts
+ library(morpheus)
+ K <- fargs$optargs$K
+ op <- optimParams(K, fargs$optargs$link, fargs$optargs)
+ best_val <- Inf
+ best_par <- list()
+ for (i in 1:fargs$optargs$nstart)
+ {
+ x_init <- c(rep(1/K,K-1), as.double(i*fargs$mu), rep(0,K))
+ par <- op$run(x_init)
+ val <- op$f( op$linArgs(par) )
+ if (val < best_val)
+ {
+ best_par <- par
+ best_val <- val
+ }
+ }
+ do.call(rbind,best_par)
+ }),
+ prepareArgs = function(fargs) {
+ library(morpheus)
+ io = generateSampleIO(fargs$n, fargs$p, fargs$beta, fargs$b, fargs$optargs$link)
+ fargs$optargs$M <- computeMoments(io$X, io$Y)
+ mu <- computeMu(io$X, io$Y, fargs$optargs)
+ fargs$mu <- mu
+ }, N=N, ncores=ncores, verbose=TRUE)
+for (i in 1:2)
+ ms[[i]] <- alignMatrices(ms[[i]], ref=rbind(p,beta,b), ls_mode="exact")
+
+ms_params <- list("N"=N, "nc"=ncores, "n"=n, "K"=K, "link"=link,
+ "p"=p, "beta"=beta, "b"=b, "nstart"=nstart)
+
+save(ms, ms_params, file="multistart.RData")
--- /dev/null
+#!/bin/bash
+
+#PBS -l nodes=1:ppn=15,mem=8gb,pmem=512mb
+#PBS -j oe
+
+#PBS -o .output
+rm -f .output
+
+WORKDIR=/workdir2/auder/morpheus/reports
+cd $WORKDIR
+
+module load R
+
+# arg --vanilla maybe possible on cluster
+for d in 10 20; do
+ for link in "logit" "probit"; do
+ R --slave --args N=1000 n=1e5 nc=15 d=$d link=$link <accuracy.R >out$d$link 2>&1
+ done
+done
--- /dev/null
+#!/bin/bash
+
+#PBS -l nodes=1:ppn=16,mem=8gb,pmem=512mb
+#PBS -j oe
+
+#PBS -o .output
+rm -f .output
+
+WORKDIR=/workdir2/auder/morpheus/reports
+cd $WORKDIR
+
+module load R
+
+# arg --vanilla maybe possible on cluster
+R --slave --args N=1000 nc=16 link=logit <timings.R >out 2>&1
--- /dev/null
+# flexmix optimization to get beta
+fmOptim <- function(X, Y, K, link)
+{
+ dat <- as.data.frame( cbind(Y,X) )
+ fm <- flexmix( cbind(Y, 1-Y) ~ .-Y, data=dat, k=K,
+ model = FLXMRglm(family = binomial(link = link)) )
+ p <- mean(fm@posterior[["scaled"]][,1])
+ out <- refit(fm)
+ beta_b <- sapply( seq_len(K), function(i) as.double( out@components[[1]][[i]][,1] ) )
+ list("p"=p, "beta"=beta_b[2:nrow(beta_b),], "b"=beta_b[1,])
+ NULL
+}
+
+# Our package optimization for beta (using mu as a starting point)
+ourOptim <- function(X, Y, K, link)
+{
+ M <- computeMoments(X, Y)
+ mu <- computeMu(X, Y, list(K=K,M=M))
+ x_init = c(1/2, as.double(mu), c(0,0))
+ optimParams(K, link, list(M=M))$run(x_init)
+ NULL
+}
+
+# Get timings for both methods with the same beta matrix
+getTimings <- function(link)
+{
+ timings <- list('fm'=matrix(0,nrow=10,ncol=7),'our'=matrix(0,nrow=10,ncol=7))
+ K <- 2
+ for (d in c(2,5,10))
+ {
+ beta <- matrix(runif(d*K,min=-5,max=5),ncol=K)
+ for (logn in 4:6)
+ {
+ n <- 10^logn
+ io <- generateSampleIO(n, rep(1/K,K-1), beta, runif(K), link)
+ timings[['fm']][d,logn] <- system.time(fmOptim(io$X,io$Y,K,link))[3]
+ timings[['our']][d,logn] <- system.time(ourOptim(io$X,io$Y,K,link))[3]
+ }
+ }
+ timings
+}
+
+#model = binomial
+link <- "logit"
+ncores <- 1
+N <- 100
+
+cmd_args <- commandArgs()
+for (arg in cmd_args)
+{
+ if (substr(arg,1,1)!='-')
+ {
+ spl <- strsplit(arg,'=')[[1]]
+ if (spl[1] == "link") {
+ link <- spl[2]
+ } else if (spl[1] == "nc") {
+ ncores <- as.integer(spl[2])
+ } else if (spl[1] == "N") {
+ N <- as.integer(spl[2])
+ }
+ }
+}
+
+library(morpheus)
+library(flexmix)
+source("../patch_Bettina/FLXMRglm.R")
+
+tm <-
+ if (ncores == 1) {
+ lapply(1:N, function(i) {
+ print(paste("Run",i))
+ getTimings(link)
+ })
+ } else {
+ library(parallel)
+ mclapply(1:N, function(i) {
+ print(paste("Run",i))
+ getTimings(link)
+ },
+ mc.preschedule=FALSE, mc.cores=ncores)
+ }
+tm_params <- list("link"=link, "N"=N, "nc"=ncores)
+
+save("tm", "tm_params", file="timings.RData")
projects / morpheus.git / commitdiff