Matlab and R

This page provides a bit of information on making the transition from Matlab to R (or vice versa), together with a few comparisons with other languages. Each langauge or system has its own strengths and weaknesses; no single system is best for everything. Needless to say, this page makes no effort whatever to include all features of these systems; see the links for further information, details, formal definitions, etc. This page contains only introductory material; think of this as your traveler's phrasebook, rather than as a reference grammar.

Both Matlab and R can be used interactively, or in batch mode. Interactively, the R system prompt is >; the Matlab system prompt on Solaris is >>; and the Scilab system prompt is -->. In what follows, I will sometimes include the system prompt and sometimes not; it will be clear from the context.

Comment

Assignment

-->a=7
 a  =

    7.

> a <- 7
> a
[1] 7

Simple Arithmetic

-->(8-3)*7+9
 ans  =
   
    44.

> (8-3)*7+9
[1] 44

Elementary Operations with Strings

Matlab requires a single quote to delimit strings; here is an example using Matlab 5 on Solaris:

>> "aa"
??? "
    |
Use single quote character instead of double quote or backward quote.

>> 'aa'

ans =

aa

Here are illustrative examples of string concatenation in Matlab, Scilab, and R. Here, we will prepend a full path to a file name. In Matlab:

>> strcat('/usr/myname/project/','thisfile.m')

ans =

/usr/myname/project/thisfile.m

-->'/usr/myname/project/' + 'thisfile.m'

 ans =

 /usr/myname/project/thisfile.m

> paste("/usr/myname/project/", "thisfile.r", sep="")
[1] "/usr/myname/project/thisfile.r"

Finally, Scilab has a strcat() function, but its behavior is somewhat different from the Matlab strcat() function.

Conversion to and from strings

Uniform Random Variables

a=unifrnd(0,1)

 a <- runif()

Comparison

Boolean Values

>> aa = (5>4)
aa =

     1

> aa <- (5>4)
> aa
[1] TRUE

> aa <- (5>4)
> aa
[1] T

Scalar and elementwise boolean operations

> # R demonstration
> x1 <- 3
> x2 <- 8
> y1 <- 9
> y2 <- 10
> xor(x1>x2, y1<=y2)
[1] TRUE

> # R demonstration
> c(TRUE,TRUE,FALSE,FALSE) & c(TRUE,FALSE,TRUE,FALSE)
> [1] TRUE FALSE FALSE FALSE
> c(TRUE,TRUE,FALSE,FALSE) & c(TRUE,FALSE,TRUE,FALSE)
> [1] TRUE TRUE TRUE FALSE

> # R demonstration
> x <- 0
> # this function will let us know if its argument is evaluated
> #   it simply prints the word evaluated, then its argument, then
> #   a return -- and it returns its argument unchanged.
> sh(x)
evaluated:  0
[1] 0
> sh <- function(x){cat("evaluated: ",x,"\n");x}
> # We attempt to avoid dividing by zero by testing first
> if (x!=0 & 8/sh(x)>0) {print("yes.")}
evaluated:  0
> # note that even though x equaled zero, the second expression
> #  8/sh(x) was evaluated anyway.
> if (x!=0 && 8/sh(x)>0) {print("yes.")}
>
> # observe that nothing was printed.  Because the system knows
> #   that the conjunction must be false once the first term is
> #   known to be false, the second expression is not even 
> #   evaluated.

Vector boolean operations

> any(c(TRUE,TRUE,FALSE))
[1] TRUE
> any(c(FALSE,FALSE,FALSE,FALSE))
[1] FALSE
> all(c(TRUE,TRUE,FALSE))
[1] FALSE
> any(c(TRUE,TRUE,TRUE,TRUE))
[1] TRUE
> x <- c(0.5,1,2,3)
> any(x<1)
[1] TRUE
> all(x<1)
[1] FALSE

Interactive Input

input('a string: ')

scan()

n = input('enter n: ')

n<-scan()
#use control-D to terminate

Vectors and Matrices

-->a=[1 2 3]

-->a=[1,2,3]

> a <- c(1,2,3)

-->a=[1 2 3]

-->b=[a 4]

> a <- c(1,2,3)

> b <- c(a,4)

Concept	Matlab/Scilab	R/S/S-Plus
Create Row Vector	[ ]	matrix()
Create row vector of 1, 2, and 3.	-->a=[1 2 3] or -->a=[1,2,3]	> a <- matrix(c(1,2,3),nrow=1)

Concept	Matlab/Scilab	R/S/S-Plus
Create Column Vector	[ ]	matrix()
Create column vector of 1, 2, and 3.	-->a=[1; 2; 3]	> a <- matrix(c(1,2,3),ncol=1)

In R, use the matrix() function with various options to construct matrices.

Concept	Matlab/Scilab	R/S/S-Plus
Create Matrix	[ ]	matrix(), rbind(), cbind()
Create a matrix with 1, 2, and 3 in the first row and 4, 5, and 6 in the second row.	-->a=[1 2 3; 4 5 6]	> a <- matrix(c(1,4,2,5,3,6),nrow=2) or > a <- rbind(c(1,2,3),c(4,5,6)) or > a <- cbind(c(1,4),c(2,5),c(3,6))

In R, the functions rbind() (row bind) and cbind() (column bind) help construct matrices.

Concept	Matlab/Scilab	R/S/S-Plus
Matrix Multiplication	*	%*%
Multiply two matrices	-->a=[1 2 3; 4 5 6] -->b=[4 5;6 7;8 9] -->a*b	> a <- rbind(c(1,2,3),c(4,5,6)) > b <- rbind(c(4,5),c(6,7),c(8,9)) > a %*% b

Matlab/Scilab uses * for multiplication of matrices; R/S use %*% for multiplication of matrices.

Concept	Matlab/Scilab	R/S/S-Plus
Elementwise (Hadamard) Matrix Multiplication	.*	*
Elementwise product of two matrices	-->a=[1 2 3; 4 5 6] -->b=[4 5 6; 7 8 9] -->a .* b	> a <- rbind(c(1,2,3),c(4,5,6)) > b <- rbind(c(4,5,6),c(7,8,9)) > a * b

Matlab/Scilab use the special symbol .* for elementwise multiplication.

Concept	Matlab/Scilab	R/S/S-Plus
Scalar Product	*	*
Multiply a vector (or matrix) by the constant 2	-->a=[1 2 3; 4 5 6] -->a * 2	> a <- rbind(c(1,2,3),c(4,5,6)) > a * 2

Scalar multiplication is straightforward in both languages.

Concept	Matlab/Scilab	R/S/S-Plus
Matrix of zeros	zeros()	matrix()
Generate a 3 by 4 matrix of zeros	-->zeros(3,4)	> a <- matrix(0,nrow=3,ncol=4)

Matlab/Scilab provide a special function for creating a matrix of zeros; R/S use the general constructor matrix().

Concept	Matlab/Scilab	R/S/S-Plus
Add a constant to every element of a vector or matrix	+	+
Demonstrate adding a constant	-->a=[1 2 3; 4 5 6];a+1	> a <- cbind(c(1,4),c(2,5),c(3,6));a+1

Adding a constant to every element of a vector or matrix is simple in both languages.

Concept	Matlab/Scilab	R/S/S-Plus
Combination of vectors into matrices	[]	rbind(), cbind()
Demonstrate combining vectors into matrices by stacking row vectors	-->a=[1 2 3] b=[4 5 6] d=[a;b]	> a <- c(1,2,3) b <- c(4,5,6) d <- rbind(a,b)
Demonstrate combining vectors into matrices by standing column vectors	-->a=[1;2;3] b=[4;5;6] d=[a b]	> a <- c(1,2,3) b <- c(4,5,6) d <- cbind(a,b)

The Matlab/Scilab [] operator is very general and elegant; R/S use the rbind() and cbind() functions to achieve the same effect.

Concept	Matlab/Scilab	R/S/S-Plus
Add a column vector to every column of a matrix	duplicate column and add as usual (*)	+
Demonstrate adding a column vector to every column	-->a=[1 2 3; 4 5 6] -->b=[5; 10]; -->a+[b b b]	> a <- cbind(c(1,4),c(2,5),c(3,6)) > b<-c(5,10) > a+b
Demonstrate adding a row vector to every row	-->a=[1 2 3; 4 5 6] -->b=[5 10 15] -->a+[b;b]	> a <- cbind(c(1,4),c(2,5),c(3,6)) > b<-c(5,10,15) > t(t(a)+b)

R's addition operator is slightly more general than this example suggests. In R/S, note that when an expression of the form M+V (with M a matrix and V a vector), the vector V is duplicated over and over until it matches the number of elements of the matrix M, and the addition is performed to M in column order (i.e. down the first column until it is finished, then down the second column, etc.).

Concept	Matlab/Scilab	R/S/S-Plus
Identity Matrix	eye()	diag()
Create 4 by 4 identity matrix	-->a=eye(4,4)	> a <- diag(4)

R's diag function can also be used to set the diagonals of a matrix.

Concept	Matlab/Scilab	R/S/S-Plus
Transpose a matrix	'	t()
Transpose a matrix	-->a=[1 2 3;4 5 6] -->a'	> a <- rbind(c(1,2,3),c(4,5,6)) > t(a)

Concept	Scilab	R
Eigenvalues of a square matrix	spec()	eigen()
Compute eigenvalues of a matrix	-->a=[1 2 3;4 5 6;7 8 9] -->spec(a)	> a <- rbind(c(1,2,3),c(4,5,6),c(7,8,9)) > eigen(a)$values

The Scilab function spec() returns a column vectors of the eigenvalues of its argument; R's eigen() function returns a list containing both the eigenvectors and eigenvalues; select the values using the $ operator as shown (more on the $ operator later).

Concept	Matlab/Scilab	R/S
Subscripting	()	[]
Get the 3d element of a vector	-->a=[1 2 3] -->a(3)	> a <- c(1,2,3) > a[3]
Get the third element of the second row	-->a=[1 2 3;4 5 6;7 8 9] -->a(2,3)	> a<-rbind(c(1,2,3),c(4,5,6),c(7,8,9)) > a[2,3] [1] 6
Get the second row of a matrix	-->a=[1 2 3;4 5 6;7 8 9] -->a(2,:)	> a<-rbind(c(1,2,3),c(4,5,6),c(7,8,9)) > a[2,]
Get the third column of a matrix	-->a=[1 2 3;4 5 6;7 8 9] -->a(:,3)	> a<-rbind(c(1,2,3),c(4,5,6),c(7,8,9)) > a[,3]

Note the difference between the syntax for selecting a row or column between the two systems.

Concept	Scilab	R
Sequences	:, linspace()	:, seq()
Vector of elements 1 through 10	-->a=1:10	> a <- 1:10
Vector of elements 1 through 10, by 0.5	-->a=1:0.5:10	> a <- seq(1,10,by=0.5)
Vector of 12 equally spaced elements, starting at 3 and ending at 17	-->a=linspace(3,17,12)	> a <- seq(3,17,length=12)

The Matlab/Scilab : operator produces a row vector; the R/S : operator produces a vector. The Matlab/Scilab : operator may also take three arguments; the R/S : operator only takes two, since the seq() function may be used instead. The R/S seq() function has other options not shown here.

Concept	Matlab/Scilab	R/S
Subsets of a Matrix	()	[]
Select elements 3 through 5 of a vector	-->a=10:20 -->a(3:5)	> a <- 10:20 > a[3:5]
Select the second, seventh, and third elements of a vector	-->a=10:20 -->a([2,7,3])	> a <- 10:20 > a[c(2,7,3)]
Select the second, seventh, and third columns of a matrix	-->a=[10:20;0:2:20] -->u(:,[2,7,3])	> a <- rbind(10:20,seq(0,20,by=2)) > a[,c(2,7,3)]
Select the second, seventh, and third rows of a matrix	-->a=[10:20;0:2:20]' -->u([2,7,3],:)	> a <- cbind(10:20,seq(0,20,by=2)) > a[c(2,7,3),]
Select everything but the second element of a vector	-->a=10:14 -->working...	> a <- 10:14 > a[-2] [1] 10 12 13 14

Note the use of the negative subscript in R/S to denote exclusion.

Concept	Scilab	R
Matrix Inversion	inv	solve
Invert a square matrix	-->a=[1,2;3,4] -->inv(a)	> a <- rbind(c(1,2),c(3,4)) > solve(a)

Decision (Alternation, If)

value = 5*(1:100)
n = input('enter n: ')
if  n>100|n<1
  'error; n out of range 1-100'
else
  choice = value(n)
end

value <- 5*(1:100)
n <- scan()
if ( n>100 | n<1 ) {
  print("error; number out of range 1-100")
} else {
  choice <- value(n)
}

Counted Loop (Repetition, DO, for)

for ii = 1:10
  jj = ii*ii
  disp(jj)
end

for (ii in 1:10) {
  jj <- ii*ii
  print(jj)
}

aa = zeros(4,4)
for ii = 1:4
  for jj = 1:4
    aa(ii,jj)=ii*jj
  end
end

aa <- matrix(0,nrow=4,ncol=4)
for (ii in 1:4) {
  for (jj in 1:4) {
    aa[ii,jj] <- ii*jj
  }
}

Here is a slightly more general example of the for loop in R. Here, we will loop through a series of (fictitious) machine names, and append berkeley.edu to each of them:

for (machine in c("orion","andromeda","centarus","cassiopeia")) {
  print(paste(machine,".berkeley.edu",sep=""))
}

While Loop

uu = unifrnd(0,1)
while uu<0.2
  uu = unifrnd(0,1)
end

uu <- runif()
while (uu<0.2) {
  uu <- runif()
}

Functions

Concept	Matlab/Scilab	R
Function Definition	write a function M-file	function()
Define a function to return the sum and product of its two arguments function name formal parameters formal return values	function [w1,w2] = testfn(aa,bb) w1 = aa + bb w2 = aa * bb	testfn <- function(aa,bb) { list(aa+bb,aa*bb) }

Consult the "Local Guide" to make sure you are loading the definitions correctly in Matlab; in Matlab 5 on Solaris, it is sufficient to place the M-file in the working directory and Matlab can find the function when it is called, and it is the file name that determines the name that is used to call the function. In Scilab the function getf can be used to explicitly load the function. In R, the definition is simply an assignment statement and can be in some R text file which must be sourced in; interactive definition is also possible. There is no necessary connection between a file name and a function definition in R; many functions can be defined in the same file.

Observe that Matlab supports multiple return value syntax; in R, you must return a composite object containing all the objects you wish to return. In Matlab and Scilab, values are returned by assigning values to the formal return variables, as shown in the example. In R, the last expression evaluated is the value returned by the function; no formal return variables are needed.

Observe that the function definition in R is simply an assignment. In R, functions are first-class objects; they can be not only called, but also passed as arguments and returned as values from other functions. There will be more information on this below.

Concept	Matlab	R
Function call	foo(arg1,arg2,...)	foo(arg1,arg2,...)
Call a function and assign the result to a (some) variable(s) (See previous example.)	-->[var1,var2] = testfn(3,4)	> varlist <- testfn(3,4)
Call an anonymous function		> (function(a,b){a+b})(4,5) [1] 9

Notice that the Matlab example simultaneously assigned values to the variables var1 and var2; the R example returned a list containing both the desired values and assigned this to varlist.

The items in the return value object in R can be named for convenience; this is a useful idiom:

> testfn <- function(aa,bb) {
+   list(sum=aa+bb,product=aa*bb)
+ }
> val <- testfn(3,4)
> val$sum
[1] 7
> val$product
[1] 12

Also, note that a function in R need not be named. The expression function(a,b){a+b} represents the function that adds its arguments. It can be used directly in a function call expression as shown in the last line in the table, where it is applied to the two arguments 4 and 5. Anonymous functions are called lambda expressions in Lisp and some other languages.