Tests/Test Time complexity/Code/Time complexity data10^3-6 iter7.r

In lpishchagina/FPOPdim2: FPOPdim2

#############################################################################################
#############################################################################################
##                                         Test                                            ##
##                                  Time complexity.                                       ##
##                            The data without changepoints                                ##
##                                      n = 1000:1000000                                   ##
#############################################################################################
#############################################################################################
###############################
#    package installation     #
###############################
#devtools::install_github("lpishchagina/FPOPdim2")

library(FPOPdim2)

#devtools::install_github("lpishchagina/OptPartitioning2D")

library(OptPartitioning2D)

library(base)
library(rstream)
library(iterators)
library(stats)

library(foreach)
library(doParallel)
set.seed(13)

###############################
#          Cluster            #
###############################
cores = detectCores()

cl <- makeCluster(cores[1]-1) #7
registerDoParallel(cl)
nb.simu <- 7


###############################
#     Function one.simu.op    #
###############################
#The function returns the runtime of the algorithm for one simulation

#type = "null" - Optimal Partitioning
#type = "pruning" - PELT

one.simu.op <- function(n,penalty, type, func = "OptPart2D"){ 
  Data  = data_gen2D(n)
  if(type == "null"){t <- system.time(OptPart2D(Data[1,], Data[2,], penalty, type = "null"))[[1]]}
  if(type == "pruning"){t <- system.time(OptPart2D(Data[1,], Data[2,], penalty, type = "pruning"))[[1]]}
  return(t)
}

###############################
#    Function one.simu.fpop   #
###############################
#The function returns the runtime of the algorithm for one simulation

#type = 1 - FPOP: only intersection, approximation of intersection = rectangle
#type = 2 - FPOP: intersection and exclusion, approximation of intersection and exclusion = rectangle
#type = 3 - FPOP: only exclusion, approximation -  disk
one.simu.fpop  <- function(n, penalty, type, func = "FPOP2D"){
  Data  = data_gen2D(n)
  if (type == 1){t <- system.time(FPOP2D(Data[1,], Data[2,], penalty, type = 1))[[1]]}
  if (type == 2){t <- system.time(FPOP2D(Data[1,], Data[2,], penalty, type = 2))[[1]]}
  if (type == 3){t <- system.time(FPOP2D(Data[1,], Data[2,], penalty, type = 3))[[1]]}
  return(t)
}

######################################################################
#                            data length generation                  #
#                           The data without changepoints            #
######################################################################
###############################
#         data length         #
###############################
length.simu <-NULL
nb.iter <- 4
for (i in 1:nb.iter) length.simu <- c(length.simu, 10^(i+2))
###############################
#            tables           #
###############################
tab.OP <- data.frame(matrix(0, nb.iter, nb.simu + 1))
colnames(tab.OP) <- c("n", paste0("Rep",1:nb.simu))

tab.PELT <- data.frame(matrix(0, nb.iter, nb.simu + 1))
colnames(tab.PELT) <- c("n", paste0("Rep",1:nb.simu))

tab.FPOP1 <- data.frame(matrix(0, nb.iter, nb.simu + 1))
colnames(tab.FPOP1) <- c("n", paste0("Rep",1:nb.simu))

tab.FPOP2 <- data.frame(matrix(0, nb.iter, nb.simu + 1))
colnames(tab.FPOP2) <- c("n", paste0("Rep",1:nb.simu))

tab.FPOP3 <- data.frame(matrix(0, nb.iter, nb.simu + 1))
colnames(tab.FPOP3) <- c("n", paste0("Rep",1:nb.simu))
###############################
#       tables filling        #
###############################
sigma <- 1
res.nb.simu <- NULL

###############################
#    tables filling FPOP2     #
###############################
for(i in 1:length(length.simu)){
  n <- length.simu[i]
  penalty <- 4*log(n)
  res.nb.simu <- foreach(i = 1:nb.simu, .combine = c, .packages=c("rstream", "FPOPdim2", "OptPartitioning2D")) %dopar% {one.simu.fpop(n, penalty = penalty, type = 2, func = "FPOP2D")}
  tab.FPOP2[i,] <- c(n, res.nb.simu)
}
###############################
#    time complexity FPOP2    #
###############################
mean.FPOP2 <- rowMeans(tab.FPOP2[,-1])
write.table(tab.FPOP2, "Table Time complexity FPOP2 10^3-6 iter7.txt" )
write.table(mean.FPOP2, "Time complexity FPOP2 10^3-6 iter7.txt", row.names = FALSE, col.names = FALSE  )
###############################
#  Plot:time complexity FPOP2 #
###############################
png(filename = "Plot log10 Time complexity FPOP2 10^3-6 iter7.png",  width = 1500, height = 1000)
  plot(log10(length.simu), log10(mean.FPOP2), xlab = "log10(data length)", ylab = "log10(mean time in second)",  main = "FPOP2:Time complexity", col = "royalblue3")
  lines(log10(length.simu), log10(mean.FPOP2), col = "royalblue3", lwd = 3)
dev.off()

png(filename = "Plot Time complexity FPOP2 10^3-6 iter7.png",  width = 1500, height = 1000)
plot(length.simu, mean.FPOP2, xlab = "data length", ylab = "mean time in second",  main = "FPOP2:Time complexity", col = "royalblue3")
lines(length.simu, mean.FPOP2, col = "royalblue3", lwd = 3)
dev.off()

###############################
#    tables filling OP        #
###############################
for(i in 1:length(length.simu)){
  n <- length.simu[i]
  penalty <- 2*log(n)
  res.nb.simu <- foreach(i = 1:nb.simu, .combine = c, .packages=c("rstream","FPOPdim2","OptPartitioning2D")) %dopar% {one.simu.op(n, penalty, type = "null", func = "OptPart2D")}
  tab.OP[i,] <- c(n, res.nb.simu) 
}

###############################
#    time complexity OP       #
###############################
mean.OP <- rowMeans(tab.OP[,-1])
write.table(tab.OP, "Table Time complexity OP 10^3-6 iter7.txt" )
write.table(mean.OP, "Time complexity OP 10^3-6 iter7.txt",  row.names = FALSE, col.names = FALSE)
###############################
#  Plot:time complexity OP    #
###############################
png(filename = "Plot log10 Time complexity OP 10^3-6 iter7.png",  width = 1500, height = 1000)
  plot(log10(length.simu), log10(mean.OP), xlab = "lg(data length)", ylab = "lg(mean time in second)",  main = "OP: Time complexity", col = "palevioletred3")
  lines(log10(length.simu), log10(mean.OP), col = "palevioletred3", lwd = 3)
dev.off()



###############################
#    tables filling PELT      #
###############################
for(i in 1:length(length.simu)){
  n <- length.simu[i]
  penalty <- 2*log(n)
  res.nb.simu <- foreach(i = 1:nb.simu, .combine = c, .packages = c("rstream", "FPOPdim2", "OptPartitioning2D")) %dopar% {one.simu.op(n, penalty, type = "pruning", func = "OptPart2D")}
  tab.PELT[i,] <- c(n, res.nb.simu) 
}
###############################
#    time complexity PELT     #
###############################
mean.PELT <- rowMeans(tab.PELT[,-1])
write.table(tab.PELT, "Table Time complexity PELT 10^3-6 iter7.txt" )
write.table(mean.PELT, "Time complexity PELT 10^3-6 iter7.txt", row.names = FALSE, col.names = FALSE  )
###############################
#  Plot:time complexity PELT  #
###############################
png(filename = "Plot log10 Time complexity PELT 10^3-6 iter7.png",  width = 1500, height = 1000)
  plot(log10(length.simu), log10(mean.PELT), xlab = "lg(data length)", ylab = "lg(mean time in second)",  main = "PELT: Time complexity", col = "orchid3")
  lines(log10(length.simu), log10(mean.PELT), col = "orchid3", lwd = 3)
dev.off()

###############################
#    tables filling FPOP3     #
###############################
for(i in 1:length(length.simu)){
  n <- length.simu[i]
  penalty <- 4*log(n)
  res.nb.simu <- foreach(i = 1:nb.simu, .combine = c, .packages=c("rstream","FPOPdim2","OptPartitioning2D")) %dopar% {one.simu.fpop(n, penalty, type = 3, func = "FPOP2D")}
  tab.FPOP3[i,] <- c(n, res.nb.simu) 
}
###############################
#    time complexity FPOP3    #
###############################
mean.FPOP3 <- rowMeans(tab.FPOP3[,-1])
write.table(tab.FPOP3, "Table Time complexity FPOP3 10^3-6 iter7.txt" )
write.table(mean.FPOP3, "Time complexity FPOP3 10^3-6 iter7.txt", row.names = FALSE, col.names = FALSE)
###############################
#  Plot:time complexity FPOP3 #
###############################
png(filename = "Plot log10 Time complexity FPOP3 10^3-6 iter7.png",  width = 1500, height = 1000)
  plot(log10(length.simu), log10(mean.FPOP3), xlab = "lg(data length)", ylab = "lg(mean time in second)",  main = "FPOP3: Time complexity", col = "purple3")
  lines(log10(length.simu), log10(mean.FPOP3), col="purple3", lwd = 3)
dev.off()

png(filename = "Plot Time complexity FPOP3 10^3-6 iter7.png",  width = 1500, height = 1000)
plot(length.simu, mean.FPOP3, xlab = "data length", ylab = "mean time in second",  main = "FPOP3:Time complexity", col = "purple3")
lines(length.simu, mean.FPOP3, col = "purple3", lwd = 3)
dev.off()



###############################
#    tables filling FPOP1     #
###############################
for(i in 1:length(length.simu)){
  n <- length.simu[i]
  penalty <- 2*log(n)
  res.nb.simu <- foreach(i = 1:nb.simu, .combine = c, .packages=c("rstream", "FPOPdim2", "OptPartitioning2D")) %dopar% {one.simu.fpop(n, penalty, type = 1, func = "FPOP2D")}
  tab.FPOP1[i,] <- c(n, res.nb.simu) 
}

###############################
#    time complexity FPOP1    #
###############################
mean.FPOP1 <- rowMeans(tab.FPOP1[,-1])
write.table(tab.FPOP1, "Table Time complexity FPOP1 10^3-6 iter7.txt" )
write.table(mean.FPOP1, "Time complexity FPOP1 10^3-6 iter7.txt", row.names = FALSE, col.names = FALSE)
###############################
#  Plot:time complexity FPOP1 #
###############################
png(filename = "Plot log10 Time complexity FPOP1 10^3-6 iter7.png",  width = 1500, height = 1000)
  plot(log10(length.simu), log10(mean.FPOP1), xlab = "lg(data length)", ylab = "lg(mean time in second)",  main = "FPOP1: Time complexity", col = "skyblue3")
  lines(log10(length.simu), log10(mean.FPOP1), col = "skyblue3", lwd = 3)
dev.off()


png(filename = "Plot Time complexity FPOP1 10^3-6 iter7.png",  width = 1500, height = 1000)
plot(length.simu, mean.FPOP1, xlab = "data length", ylab = "mean time in second",  main = "FPOP1:Time complexity", col = "skyblue3")
lines(length.simu, mean.FPOP1, col = "skyblue3", lwd = 3)
dev.off()


###############################
#  Plot:time complexity       #
###############################

min_lgy = min(log10(mean.FPOP1),log10(mean.FPOP2), log10(mean.FPOP3), log10(mean.OP),log10(mean.PELT))
max_lgy = max(log10(mean.FPOP1),log10(mean.FPOP2), log10(mean.FPOP3), log10(mean.OP),log10(mean.PELT)) 

min_lgx = min(log10(length.simu))
max_lgx = max(log10(length.simu))

png(filename = "Plot log10 Time complexity Comparison 10^3-6 iter7.png",  width = 1500, height = 1000)

  plot(log10(length.simu), log10(mean.FPOP1), col = "skyblue3", xlab = "lg(data length)", ylab = "lg(mean time in second)",  main = "Time complexity", ylim = c(min_lgy, max_lgy), xlim = c(min_lgx, max_lgx))
  lines(log10(length.simu), log10(mean.FPOP1), col="skyblue3", lwd = 3)

  points(log10(length.simu), log10(mean.FPOP3), col = "purple3")
  lines(log10(length.simu), log10(mean.FPOP3), col="purple3", lwd = 3)

  points(log10(length.simu), log10(mean.OP), col = "palevioletred3")
  lines(log10(length.simu), log10(mean.OP), col="palevioletred3", lwd = 3)

  points(log10(length.simu), log10(mean.PELT), col = "orchid3")
  lines(log10(length.simu), log10(mean.PELT), col= "orchid3", lwd = 3)

  points(log10(length.simu), log10(mean.FPOP2), col = "royalblue3")
  lines(log10(length.simu), log10(mean.FPOP2), col="royalblue3", lwd = 3)
  ############################### 
  #       legend                #
  ###############################
  location = "topleft"
  labels = c("OP","PELT", "FPOP1", "FPOP2", "FPOP3")
  colors = c("palevioletred3", "orchid3","skyblue3","royalblue3", "purple3")
  legend(location, labels, fill = colors)

dev.off()


############################### 
#  stop cluster               #
###############################

#stop cluster
stopCluster(cl)