Introduction to F

Fortran (FORmula TRANslation) was introduced in 1957 and remains the language of choice for most scientific programming. The latest standard, Fortran 90, includes extensions that are familiar to users of C. Some of the most important features of Fortran 90 include recursive subroutines, dynamic storage allocation and pointers, user defined data structures, modules, and the ability to manipulate entire arrays.

Fortran 90 is compatible with Fortran 77 and includes syntax that is no longer considered desirable. F is a subset of Fortran 90 that includes only its modern features, is compact, and is easy to learn.

1. Introduction 2. Do construct 3. If construct 4. Formatted output
5. Files 6. Arrays 7. Allocate 8. Random number sequences
9. Recursion 10. Character variables 11. Bit manipulation (not done) 12. Complex variables
13. Pointers (not done) 14. References 15. Links

Instructions for running F on Unix workstations and on the Power Macintosh are available.

1. Introduction
In the following, we will assume that the reader is familiar with a programming language such as True BASIC. We first give an example of an F program.

program product_example
   real :: m, a, force
   m = 2.0                       ! mass in kilograms
   a = 4.0                       ! acceleration in mks units
   force = m*a                   ! force in newtons
print *, force
end program product_example
This program is similar to Program product in Chapter 2 of Gould and Tobochnik. The features of F included in the above program include:

The first statement must be a program statement; the last statement must have a corresponding end program statement.

Integer numerical variables and floating point numerical variables are distinguished. The names of all variables must be between 1 and 31 alphanumeric characters of which the first must be a letter and the last must not be an underscore.

The types of all variables must be declared.

Real numbers are written as 2.0 rather than 2.

The case is significant, but two names that differ only in the case of one or more letters cannot be used together. All keywords (words that are part of the language and cannot be redefined.) are written in lower case. Some names such as product are reserved and cannot be used as names.

Comments begin with a ! and can be included anywhere in the program.

Statements are written on lines which may contain up to 132 characters.

The asterisk (*) following print is the default format.

We next introduce syntax that allows us to enter the desired values of m and a from the keyboard. Note the use of the (unformatted) read statement and how character strings are printed.

program product_example2
   real :: m, a, force
   ! SI units
   print *, "mass m = ?"
   read *, m
   print *, "acceleration a = ?"
   read *, a
   force = m*a
   print *, "force (in newtons) =", force
end program product_example2
2. Do construct
F uses a do construct to have the computer execute the same statements more than once. An example of a do construct follows:
program series
   integer :: n
   real :: sum_series       ! sum is a keyword
   sum_series = 0.0
!  add the first 100 terms of a simple series
   do n = 1, 100
      sum_series = sum_series + 1.0/real(n)**2
      print *, n,sum_series
   end do
end program series
Note that n is an integer variable. In this case the do statement specifies the first and last values of n; n increases by unity (default). The block of statements inside a loop is indented for clarity.

Because the product n*n is done using integer arithmetic, it is better to convert n to a real variable before the multiplication is done. Also note that exponentiation is done using the operator **.

3. If construct
In the next program example, the do loop is exited by satisfying a test.

program series_test
!  illustrate use of do construct
   integer :: n
!  choose large value for relative change
   real :: sum_series, newterm, relative_change
   n = 0
   sum_series = 0.0
   do
      n = n + 1
      newterm = 1.0/(n*n)
      sum_series = sum_series + newterm
      relative_change = newterm/sum_series
      if (relative_change < 0.0001) then
         exit
      end if
      print *, n,relative_change,sum_series
   end do
end program series_test
The new features of F included in the above program include:

A do construct can be exited by using the exit statement.

The if construct allows the execution of a sequence of statements (a block) to depend on a condition. The if construct is a compound statement and begins with if ... then and ends with end if. The block inside the if construct is indented for clarity. Examples of more general if constructs using else and else if statements are given in Program test_factorial.

relation operator
less than <
less than or equal <=
equal ==
not equal /=
greater than >
greater than or equal >=
Table 1. Summary of relational operators.

The following program illustrates the use of a kind parameter and a named do construct:

program series_example
!  illustrate use of kind parameter and named do loop
   integer, parameter :: double = 8
   integer :: n
   real (kind = double) :: sum_series, newterm, relative_change
   n = 0
   sum_series = 0.0
   print_change: do
      n = n + 1
      newterm = 1.0/real(n, kind = double)**2
      sum_series = sum_series + newterm
      relative_change = newterm/sum_series
      if (relative_change < 0.0001) then
         exit print_change
      end if
      print *, n,relative_change,sum_series
   end do print_change
end program series_example

Variables may have a particular hardware representation such as double precision by using the kind = double in parenthesis after the keyword real representing the data type. A more general use of the parameter statement is given in Program drag.

Double precision in Fortran 90 on the SGI and in F on the Power Macintosh is done by letting

double = 8
However double precision in F on the SGI is done by letting
double = 2

The do and end do statements must either have the same name or both be unnamed. In general, a do construct is named to make explicit which do construct is exited in the case of nested do constructs. The use of a named do construct in the above example is unnecessary and is for illustrative purposes only.

Subprograms are called from the main program or other subprograms. As an example, the following program adds and multiplies two numbers that are inputed from the keyboard. Note that the variables x and y are public and are available to the main program.

module common
   public :: initial,add,multiply
   integer, parameter, public :: double = 8
   real (kind = double), public :: x,y

contains

subroutine initial()
   print *, "x = ?"
   read *,x
   print *, "y = ?"
   read *,y
end subroutine initial

subroutine add(sum2)
   real (kind = double), intent (in out) :: sum2
   sum2 = x + y
end subroutine add

subroutine multiply(product2)
   real (kind = double), intent (in out) :: product2
   product2 = x*y
end subroutine multiply

end module common

program tasks            ! illustrate use of module and subroutines
! note how variables are passed
   use common

   real (kind = double) :: sum2, product2
   call initial()                 ! initialize variables
   call add(sum2)                 ! add two variables
   call multiply(product2)
   print *, "sum =", sum2, "product =", product2
end program tasks

Subprograms (subroutines and functions) are contained in modules. The form of a module, subroutine, and a function is similar to that of a main program.

A module is accessed in the main program by the use statement.

Subroutines are invoked in the main program by using the call statement.

A subprogram always has access to other entities in the module.

The subprograms in a module are preceded by a contains statement.

Variables and subprograms may be declared public in a module and be available to the main program (and other modules).

Information can also be passed as an argument to each subprogram as are the variables sum2 and product2. A open parenthesis () is needed even if there are no arguments. The intent of each dummy argument of a program must be indicated.

The module(s) can be a separate file.

4. Formatted output
The structure of
Program cool is similar to Program tasks. Note the use of the modulo function and the use of format specifications. In place of the asterisk * denoting the default format, we have used a format specification which is a list of edit descriptors. An example from Program cool is

print "(t7,a,t16,a,t28,a)", "time","T_coffee","T_coffee - T_room"
The t (tab) edit descriptor is used to skip to a specified position on an output line. The edit descriptor a (alphanumeric) is for character strings. An example of the f (floating point) descriptor is given by
print "(f10.2,2f13.4)",t,T_coffee,T_coffee - T_room
The edit descriptor f13.4 means that a total of thirteen positions are reserved for printing a real value rounded to 4 places after the decimal point. (The decimal point and a minus sign occupy two out of the thirteen positions.) The edit descriptor 2f13.4 means that the edit descriptor f13.4 is done twice. The other common edit descriptor is i (integer).

Comment on Program drag
The only new syntax in Program drag is the use of the parameter statement:

real (kind = double), public, parameter :: g = 9.8
A parameter is a named constant. The value of a parameter is fixed by its declaration and cannot be changed during the execution of a program.

5. Files
Program save_data illustrates how to open a new file, write data in a file, close a file, and read data from an existing file.

program save_data
!  illustrate writing and reading file
   integer :: i,j,x
   character(len = 32) :: file_name
   print *, "name of file?"
   read *, file_name
   open (unit=5,file=file_name,action="write",status="new")
   do i = 1,4
      x = i*i
      write (unit=5,fmt=*) i,x
   end do
   close(unit=5)
   open (unit=1,file=file_name,action="read",status="old")
   do i = 1,4
      read (unit=1,fmt = *) j,x
      print *, j,x
   end do
   close(unit=1)
end program save_data
Input/output statements refer to a particular file by specifying its unit. The read and write statements do not refer to a file directly, but refer to a file number which must be connected to a file. There are many variations on the open statement, but the above example is typical. The values of the action specifier are read, write, and readwrite (default). Values for status are old, new, replace, or scratch.

If we plan to reuse data on the same system with the same compiler, we can use unformatted input/output to save the overhead, extra space, and the roundoff error associated with the conversion of the internal representation of a value to its external representation. Of course, the latter is machine and compiler dependent. Unformatted access is very useful when data is generated by one program and then analyzed by a separate program on the same computer. To generate unformatted files, omit the format specification. Examples of programs which use direct access and records are available.

6. Arrays
The definition and use of arrays is illustrated in
Program vector.

module common
   public :: initial,cross

   contains

   subroutine initial(a,b)
      real, dimension (:), intent(out) :: a,b
      a(1:3) = (/ 2.0, -3.0, -4.0 /)
      b(1:3) = (/ 6.0, 5.0, 1.0 /)
   end subroutine initial

   subroutine cross(r,s)
      real, dimension (:), intent(in) :: r,s
      real, dimension (3) :: cross_product
      ! note use of dummy variables
      integer :: component,i,j
      do component = 1,3
         i = modulo(component,3) + 1
         j = modulo(i,3) + 1
         cross_product(component) = r(i)*s(j) - s(i)*r(j)
      end do
      print *, ""
      print *, "three components of the vector product:"
      print "(a,t10,a,t16,a)", "x","y","z"
      print *, cross_product
   end subroutine cross

   end module common

   program vector            ! illustrate use of arrays
      use common
      real, dimension (3) :: a,b
      real :: dot
      call initial(a,b)
      dot = dot_product(a,b)
      print *, "dot product = ", dot
      call cross(a,b)
   end program vector
The main features of arrays include:

An array is declared in the declaration section of a program, module, or procedure using the dimension attribute. Examples include

real, dimension (10) :: x,y
real, dimension (1:10) :: x,y
integer, dimension (-10:10) :: prob
integer, dimension (10,10) :: spin

The default value of the lower bound of an array is 1. For this reason the first two statements are equivalent to the first.

The lower bound of an array can be negative.

The last statement is an example of two-dimensional array.

Rather than assigning each array element explicitly, we can use an array constructor to give an array a set of values. An array constructor is a one-dimensional a list of values, separated by commas, and delimited by "(/" and "/)". An example is

a(1:3) = (/ 2.0, -3.0, -4.0 /)
is equivalent to the separate assignments
a(1) = 2.0
a(2) = -3.0
a(3) = -4.0

Note that the array cross_product can be referenced by one statement:

print *, cross_product

F has many vector and matrix multiplication functions. For example, the function dot_function operates on two vectors and returns their scalar product. Some useful array reduction functions are maxval, minval, product, and sum.

7. Allocate statement
One of the better features of Fortran 90 is dynamic storage allocation. That is, the size of an array can be changed during the execution of the program. The use of the allocate and deallocate statements are illustrated in the following. Note the use of the implied do loop.

program dynamic_array
!  simple example of dynamic arrays
   real, dimension (:), allocatable :: x
   integer :: i,N
   
   N = 2
   allocate(x(N:2*N))
!  implied do loop
   x(N:2*N) = (/ (i*i, i = N, 2*N) /)
   print *, x
   deallocate(x)
   allocate(x(N:3*N))
   x = (/ (i*i, i = N, 3*N) /)
   print *, x
end program dynamic_array
An example of passing arrays:
module param

integer, public, parameter :: double = 8

end module param

module common

use param
private

public :: initial

integer, public :: N

contains

subroutine initial(x)
   real (kind = double), intent(inout), dimension(:) :: x
   N = 100
   x(1) = 1.0
end subroutine initial

end module common

program test
   use param
   use common
   real (kind = double), allocatable,dimension (:) :: x
   N = 10
   allocate(x(N))
   call initial(x)
end program test

8. Random number sequences
Fortran 90 includes several useful built-in procedures. One of the most useful ones is subroutine random_number. Although it is a good idea to write your own random number generator using an algorithm that you have tested on the particular problem of interest, it is convenient to use subroutine random_number when you are debugging your program or if accuracy is not important. The following program illustrates several uses of subroutine random_number and random_seed. Note that the argument rnd of random_number must be real, has intent out, and can be either a scalar or an array.

program random_example
   real :: rnd
   real, dimension (:), allocatable :: x
   integer, dimension(2) :: seed, seed_old
   integer :: L,i,n_min,n_max,ran_int,sizer
!  generate random integers between n_min and n_max
!  dimension of seed is one in F and two in Fortran 90.
   call random_seed(sizer)
   print *, sizer
!  illustrate use of put and get
   seed(1) = 1239
   seed(2) = 9863        ! need for Fortran 90
   call random_seed(put=seed)
   call random_seed(get=seed_old)
!  confirm value of seed
   print *, "seed = ", seed_old
   L = 100         ! length of sequence
   n_min = 1
   n_max = 10
   do i = 1,L
      call random_number(rnd)
      ran_int = (n_max - n_min + 1)*rnd + n_min
      print *,ran_int
   end do
!  assign random numbers to array x as another example
   allocate(x(L))
   call random_number(x)
   print "(4f13.6)", x
   call random_seed(get=seed_old)
   print *, "seed = ", seed_old
end program random_example
Note how subroutine random_seed is used to specify the seed. This specification is useful when the same random number sequence is used to test a program.

9. Recursion
A simple example of a recursive definition is the factorial function

factorial(n) = n! = n(n-1)(n-2) ... 1
A recursive definition of the factorial is
factorial(1) = 1 factorial(n) = n factorial(n-1)
A program that closely parallels the above definition follows. Note how the word recursive is used.
module fact

public :: f
contains

recursive function f(n) result (factorial_result)
   integer, intent (in) :: n
   integer :: factorial_result

   if (n <= 0) then
      factorial_result = 1
   else
      factorial_result = n*f(n-1)
   end if
end function f

end module fact

program test_factorial
   use fact

   integer :: n
   print *, "integer n?"
   read *, n
   print "(i4, a, i10)", n, "! = ", f(n)
end program test_factorial

A less simple example (taken from pp. 98-99 in The Fun of Computing,John G. Kemeny, True BASIC (1990)) is given two integers, n and m, what is their greatest common divisor, that is, the largest integer that divides both? For example, if n = 1000 and m = 32, than the greatest common divisor (gcd) is gcd = 8.

One method for finding the gcd is to integer divide n by m. We write n = q m + r, where q is the quotient and r is the remainder. If r = 0, then m divides n and m is the gcd. Otherwise, any divisor of m and r also divides n, and hence gcd(n,m) = gcd(m,r). Because r < m, we have made progress. As an example, take n = 1024 and m = 24. Then q = 42 and r = 16. So we want gcd(24,16). Now q = 1 and r = 8 and we calculate gcd(16,8). Finally q = 2, and r = 0 so gcd = 8. The following program implements this idea.

module gcd_def

public :: gcd
contains

recursive function gcd(n,m) result (gcd_result)
   integer, intent (in) :: n,m
   integer :: gcd_result
   integer :: remainder

   remainder = modulo(n,m)
   if (remainder == 0) then
      gcd_result = m
   else
      gcd_result = gcd(m,remainder)
   end if
end function gcd

end module gcd_def

program greatest
   use gcd_def

   integer :: n,m
   print *, "enter two integers n, m"
   read *, n,m
   print "(a,i6,a,i6,a ,i6)", "gcd of",n," and",m,"=",gcd(n,m)
end program greatest

The example of recursion given in almost all introductory textbooks is the towers of Hanoi. To save space, a discussion is given elsewhere together with the program. (not finished)

The volume of a d-dimensional hypersphere of unit radius can be related to the area of a (d - 1)-dimensional hypersphere. The following program uses a recursive subroutine to integrate numerically a d-dimensional hypersphere:

module common
   public :: initialize,integrate

   integer, parameter, public :: double = 8
   real (kind = double), parameter, public :: zero = 0.0
   real (kind = double), public :: h, volume
   integer, public :: d

contains

subroutine initialize()
   print *, "dimension d?"
   read *, d                         ! spatial dimension
   print *, "integration interval h?"
   read *, h
   volume = 0.0
end subroutine initialize

recursive subroutine integrate(lower_r2, remaining_d)
!  lower_r2 is contribution to r^2 from lower dimensions
   real(kind = double),intent (in) :: lower_r2
   integer, intent (in) :: remaining_d  ! # dimensions to integrate
   real (kind = double) :: x

   x = 0.5*h   ! mid-point approximation
   if (remaining_d > 1) then
      lower_d: do
         call integrate(lower_r2 + x**2, remaining_d - 1)
         x = x + h
         if (x > 1) then
            exit lower_d
         end if
      end do lower_d
   else   
      last_d: do
         if (x**2 + lower_r2 <= 1) then
           volume = volume + h**(d - 1)*(1 - lower_r2 - x**2)**0.5
         end if
         x = x + h
         if (x > 1) then
            exit last_d
         end if
      end do last_d
   end if
end subroutine integrate

end module common

program hypersphere
!  original program by Jon Goldstein
   use common

   call initialize()
   call integrate(zero, d - 1)
   volume = (2**d)*volume         ! only considered positive octant
   print *, volume
end program hypersphere

10. Character variables
The only instrinsic operator for character expressions is the concatenation operator //. For example, the concatenation of the character constants string and beans is written as

"string"//"beans"
The result, stringbeans, may be assigned to a character variable. A useful example of concatenation is given in the following program.
program write_files
!  test program to open n files and write data
   integer :: i,n
   character(len = 15) :: file_name
   n = 11
   do i = 1,n
!     assign number.dat to file_name using write statement
      write(unit=file_name,fmt="(i2.2,a)") i,".dat" 
!     // is concatenation operator
      file_name = "config"//file_name
      open (unit=1,file=file_name,action="write",status="replace")
      write (unit=1, fmt=*) i*i,file_name
      close(unit=1)
   end do
end program write_files
Note the use of the write statement to build a character string for numeric and character components.

11. Bit manipulation

12. Complex variables
Fortran 90 is uniquely suited to handle complex variables. The following program illustrates the way complex variables are defined and used.

program complex_example
   integer, parameter :: double = 2
   real (kind = double), parameter :: pi = 3.141592654
   complex (kind = double) :: b,bstar,f,arg
   real (kind = double) :: c
   complex :: a
   integer :: d
   ! A complex constant is written as two real numbers, separated by
   ! a comma and enclosed in parentheses.
   a = (2,-3)
   ! If one of part has a kind, the other part must have same kind
   b = (0.5_double,0.8_double)
   print *, "a =", a      ! note that a has less precision than b
   print *, "a*a =", a*a
   print *, "b =", b
   print *, "a*b =", a*b
   c = real(b)     ! real part of b
   print *, "real part of b =", c
   c = aimag(b)      ! imaginary part of b
   print *, "imaginary part of b =", c
   d = int(a)
   print *, "real part of a (converted to integer) =", d
   arg = cmplx(0.0,pi)
   b = exp(arg)       ! done in two lines for ease of reading only
   bstar = conjg(b)    ! complex conjugate of b
   f = abs(b)       ! absolute value of b
   print *, "properties of b =", b,bstar,b*bstar,f
end program complex_example

13. Pointers

14. References

Walter S. Brainerd, Charles H. Goldberg, and Jeanne C. Adams, Programmer's Guide to F, Unicomp (1996).

Michael Metcalf and John Reid, The F Programming Language, Oxford University Press (1996).

15. Links

Please send comments and corrections to Harvey Gould, hgould@clarku.edu.

Updated 17 March 2001.