Change to runtime codes.

R/T_statistic_Runtimes.R

@@ -10,6 +10,7 @@ library(microbenchmark)
 library(MASS)
 library(MonteCarlo)
 library(data.table)
+library(tikzDevice)
 library(ggplot2)
 
 setwd("/home/jaroslav-borodavka/Schreibtisch/Dissertation/Arbeit/Bernoulli Submission/codes/codes_git_project/R")
@@ -18,11 +19,33 @@ setwd("/home/jaroslav-borodavka/Schreibtisch/Dissertation/Arbeit/Bernoulli Submi
 source("T_statistic_Functions.R")
 
 # benchmark tests
-
 n = 100
+
+d = 3
+benchmark_res_d3 <- microbenchmark(T_unif(n, d, 1), T_unif(n, d, 2), T_unif(n, d, 3), T_unif(n, d, 4), T_unif(n, d, 5), T_unif(n, d, 6), times=100, unit="ms")
 d = 5
+benchmark_res_d5 <- microbenchmark(T_unif(n, d, 1), T_unif(n, d, 2), T_unif(n, d, 3), T_unif(n, d, 4), T_unif(n, d, 5), T_unif(n, d, 6), times=100, unit="ms")
+
+print(benchmark_res_d3)
+print(benchmark_res_d5)
+
+# convert to milliseconds
+benchmark_res_d3$time = benchmark_res_d3$time/10^6
+benchmark_res_d5$time = benchmark_res_d5$time/10^6
+
+# concatenate data
+df_3 <- data.frame(dim = rep(3, length(benchmark_res_d3$expr)), benchmark_res_d3)
+df_5 <- data.frame(dim = rep(5, length(benchmark_res_d5$expr)), benchmark_res_d5)
+df <- rbind(df_3, df_5)
+
+# Create box plot
+p <- ggplot(df, aes(x=expr, y=time, fill = factor(dim))) + 
+  geom_boxplot(outlier.colour="black", outlier.shape=23, outlier.size=1,
+               size = 0.5,
+               width=0.5) +
+  scale_x_discrete(name = expression(beta), labels = seq(1,6)) +
+  scale_y_continuous(name = paste(expression(t), "[ms]"), breaks = seq(0, 1100, 100), limits = c(0, 1100)) +
+  scale_fill_brewer(name = "Dimension", palette="Blues")
+p
 
-res <- microbenchmark(T_unif(n, d, 1), T_unif(n, d, 2), T_unif(n, d, 3), T_unif(n, d, 4), T_unif(n, d, 5), T_unif(n, d, 6), times=100, unit="s")
-print(res)
-ggplot2::autoplot(res)
-boxplot(res)
+#ggsave("running_times_boxplot.pdf")