R basics workshopJ. Sebastián TelloIván JiménezCenter for Conservation and Sustainable DevelopmentMissouri Botanical Garden

9. Flow Control

Flow control lets you define how your scripts run

• There are a number of constructs in R that allow you to control the flow of the code

• There are mainly 3 types:

• Loops – for, while, repeat• Breaking points – break, next• Conditionals – if, else and ifelse

• We will focus on: for, while and if

• For help:

?Control

“for” loops

• A loop is the repetition of a piece of code “n” times

• for is the most common construct to create loops

• This is the general structure of a “for” loop:

for(i in v){

code…}

For each value that i takes from vector v, repeat:{

this code}

Which means:

Tab/space

• Easy example 1:

v <- 1:10

for(i in v){

print(i)}

The vector v has values from 1 to 10 every 1

For each value that i takes from vector v, repeat:print the value of i into the screen

“for” loops

Which means:

• Easy example 2:

v <- letters

v

for(i in v){

print(i)}

“for” loops

• Easy example 3:

v <- letters

length(v)

result <- 0

for(i in v){

print(i)result <- result + 1

}

result

“for” loops

v <- c(1,3,5,2,4)

result <- 0

for(i in 1:length(v)){

print( c(i, v[i]) )

result <- result + v[i]}

result

• Easy example 4:

“for” loops

col.v <- rainbow(100)cex.v <- seq(1, 10, length.out=100)

plot(0:1, 0:1, type="n")

for(i in 1:200){

print(i)

points(runif(1), runif(1), pch=16, col=sample(col.v, 1), cex=sample(cex.v, 1))

Sys.sleep(0.1)}

• Easy example 5:

“for” loops

BatData <- read.table(file=file.choose(),header=TRUE, sep="\t")

Or if the files is in your working directory:

BatData <- read.table(file="BatsEnviroAmerica.txt",header=TRUE, sep="\t")

• Open the file “BatsEnviroAmerica.txt”

“for” loops

class(BatData)

names(BatData)

rich <- BatData$richness

enviro <- BatData[,5:ncol(BatData)]

enviro[1:5, ]

• Open the file “BatsEnviroAmerica.txt”

“for” loops

LM.R2 <- rep(NA, ncol(enviro))

LM.R2

for(i in 1:ncol(enviro)){

LM.i <- lm(rich ~ enviro[,i])

res.LM.i <- summary(LM.i)

LM.R2[i] <- res.LM.i$adj.r.squared}

LM.R2

“for” loops

LM.R2

names(LM.R2) <- names(enviro)

barplot(LM.R2)

“for” loops

“while” loops

• while is sometimes also very useful

• This is the general structure of a “while” loop:

while(condition){

code…}

While this condition is TRUE, repeat:{

this code}

Which means:

• Easy example 1:

v <- 1:10for(i in v){

print(i)}

“while” loops

v <- 1:10i <- 0while(i < max(v)){

i <- i+1print(i)

}

• Easy example 1:

“while” loops

i <- 0while(i < max(v)){

i <- i+1print(i)

}

i <- 0while(i < max(v)){

print(i)i <- i+1

}

Version 1

Version 2

v <- 1:10

“while” loops

Bp <- 0.1; Dp <- 0.1; Np <- 1-Bp-Dp

max.t <- 100; time <- 0; abund <- 10

plot(c(0, max.t), c(0, 100), type="n")

while(abund>0 & time<= max.t){

change <- sample(c(-1,0,1), size=abund, prob=c(Dp, Np, Bp), replace=TRUE)

abund <- abund + sum(change)time <- time + 1

points(time, abund, pch=16, col="black")}

• Easy example 2:

• if controls the flow by allowing code to run only if a condition is met

• Easy example 1:v <- 1:10

for(i in v){

print(i)

if(i == 5)print("Reached 5")

}

“if” condition

• if controls the flow by allowing code to run only if a condition is met

• Easy example 1:v <- 1:10

for(i in v){

print(i)

if(i == 5){

print("Reached 5")break()

}}

“if” condition and “break”

trait <- 0; max.time <- 100

plot(c(0,max.time), c(-20, 20), type="n", ylab="Trait Value", xlab="Time")

points(0, trait, pch=16, col="black")

for(i in 1:max.time){

trait.shift <- rnorm(1, 0, 0.5)trait <- trait + trait.shift

if(trait.shift > 0) COL <- "gold"if(trait.shift < 0) COL <- "lightblue"

points(i, trait, pch=16, col=COL)

Sys.sleep(0.2)}

“if” condition

M <- matrix(rpois(50000000, 10), ncol=50)

M[1:5,]

dim(M)

Avoiding loops

• Loops are extremely useful, but slow. When possible, avoid them.

• Often, you will be working with large data sets. Lets simulate a matrix of 50 species abundances in 1,000,000 sites

abund.1 <- numeric()

system.time( { for(i in 1:nrow(M)) {

abund.1 <- c(abund.1, sum(M[i,])) }

})

Avoiding loops

• How to calculate the number of individuals at each site (sum by rows)?

• Option 1 – a ‘for’ lool

abund.2 <- rep(NA, nrow(M))

system.time( { for(i in 1:nrow(M)) {

abund.2[i] <- sum(M[i,]) }

})

Avoiding loops

• How to calculate the number of individuals at each site (sum by rows)?

• Option 2 – a better ‘for’ lool

system.time( { abund.3 <- apply(M, 1, sum)})

?apply

Avoiding loops

• How to calculate the number of individuals at each site (sum by rows)?

• Option 3 – use a function of the family ‘apply’

system.time( { abund.4 <- colSums(M)})

Vectorization

• How to calculate the number of individuals at each site (sum by rows)?

• Option 4 – Vectorize! Use a built-in function in R that was written in other code (e.g., C, C++, Fortran)

Exercise 9Flow Control