Fortran 77 for beginners
| IDRIS adaptation of the Fortran 77 manual from: |
|
University of Strathclyde Computer Centre Curran Building -- 100 Cathedral Street -- Glasgow (See Copyright) |
Table of contents
- 1 Computer basics
- 1.1 The memory
- 1.2 Programming languages
- 1.2.1 High level languages
- 2 Constants and variables
- 2.1 The CHARACTER type
- 2.2 The INTEGER type
- 2.3 The REAL type
- 2.4 Variables
- 2.4.1 Explicit typing
- 2.4.2 Implicit (or default) typing
- 2.4.3 Assigning a value
- 2.4.3.1 The READ statement
- 2.4.3.2 The assignment statement
- 2.4.3.3 Type rules
- 3 Arithmetic expressions and
assignments
- 4 A simple program
- 4.1 The PRINT statement
- 4.2 The PROGRAM statement
- 4.3 The END and STOP statements
- 4.4 Program layout (fixed format)
- 4.5 Comments
- 4.6 A simple program
- 5 Control structures -
Conditional execution
- 5.1 The LOGICAL type
- 5.2 Logical expressions
- 5.2.1 Relational expressions
- 5.2.2 Composite logical expressions
- 5.3 Logical assignment
- 5.4 The logical IF statement
- 5.5 The block IF structure
- 6 Control structures - Iteration
- 6.1 The GOTO statement
- 6.2 Count controlled loops
- 6.3 The DO-loop
- 6.3.1 Execution
- 6.3.2 Restrictions and other notes
- 6.3.3 The control variable
- 6.3.4 Nested DO loops
- 7 Arrays
- 7.1 Array declarations
- 7.2 Use of arrays and array elements
- 7.3 Initialising an array
- 7.3.1 The DATA statement
- 7.3.2 The implied DO list
- 7.4 Input and output of arrays
- 7.5 Multi-dimensional arrays
- 8 Input and output
- 8.1 The FORMAT statement
- 8.2 Formatted input
- 8.2.1 The I format descriptor
- 8.2.2 The F format descriptor
- 8.2.3 The E format descriptor
- 8.2.4 Repeat count
- 8.2.5 The X format descriptor
- 8.2.6 The T format descriptors
- 8.3 Formatted output
- 8.3.1 Vertical spacing
- 8.3.2 The I format descriptor
- 8.3.3 The F format descriptor
- 8.3.4 The E format descriptor
- 8.3.5 The literal format descriptors
- 8.4 More general input/output statements
- 8.5 The OPEN statement
- 8.6 Repetition of format specifications
- 8.7 Multi-record specifications
- 8.8 Sequential unformatted I/O
- 8.9 Direct access unformatted I/O
- 9 Functions ans Subroutines
- 9.1 Intrinsic functions
- 9.2 External functions
- 9.2.1 The FUNCTION statement
- 9.2.1.1 Type
- 9.2.1.2 The argument list
- 9.2.2 The function reference
- 9.2.2.1 Actual and dummy arguments
- 9.2.3 Evaluation of a function
- 9.2.4 Examples
- 9.2.1 The FUNCTION statement
- 9.3 Statement functions
- 9.3.1 Rules
- 9.4 Subroutines
- 9.5 Procedures as arguments
- 9.6 Local variables
- 10 The character Type
- 10.1 CHARACTER constants
- 10.2 CHARACTER variables
- 10.2.1 Arrays
- 10.2.2 Assignment
- 10.3 CHARACTER expressions
- 10.3.1 Concatenation
- 10.3.2 Extraction of a substring
- 10.4 Input and output
- 10.5 Logical expressions
- 11 Additional information types
- 11.1 DOUBLE PRECISION
- 11.1.1 DOUBLE PRECISION constants
- 11.1.2 DOUBLE PRECISION variables
- 11.1.3 Input and output
- 11.1.4 Expressions
- 11.1.5 Functions
- 11.2 COMPLEX
- 11.2.1 COMPLEX constants
- 11.2.2 COMPLEX variables
- 11.2.3 Input and output
- 11.2.4 COMPLEX expressions
- 11.2.5 COMPLEX functions
- 11.1 DOUBLE PRECISION
- 12 Other Fortran 77 features
- 12.1 EQUIVALENCE
- 12.2 COMMON
- 12.3 BLOCKDATA
- 12.6 Other obsolescent features
- 12.6.1 Arithmetic IF
- 12.6.2 Computed GOTO
- 13 Writing and testing programs
- 14 The standard ASCII table
- 15 Copyright
-1- Computer Basics
- 1.1 The memory
- 1.2 Programming languages
- 1.2.1 High level languages
A computer (often called simply a machine) is a device for the fast, accurate processing of symbolic information under the control of a stored sequence of instructions called a program.
Figure 1 is a schematic diagram of a simple computer system.
The control unit controls the operation of the whole machine. It:
- reads information from the input and stores it in the memory.
- fetches each program instruction in sequence from the memory, interprets it and supervises its execution.
- fetches information required for processing from the memory and sends it to the arithmetic and logic unit, which performs the required computations.
- receives the result of each computation from the arithmetic and logic unit and stores it in the memory or sends it directly to the output.
- fetches information from the memory and sends it to the output.
1.1 The memory
Figure 2: The memory
The memory stores programs and information. It consists of an array of storage units called words, all of equal length and numbered in sequence. The number of each word is called its address.
Each word contains a row of storage elements, which can individually be set to either of two states, conventionally represented as 0 and 1. Thus information is represented by binary codes.
When a program is run, each instruction in sequence is fetched from memory and executed. An instruction can order the control unit to fetch its next instruction from an address other than the next in memory. In this way, different sequences of instructions can be executed. This makes it possible to execute instruction sequences conditionally and iteratively.
1.2 Programming languages
To be executable, a program must be written in the binary machine code recognised by the processor. Machine code programming is difficult and prone to error. Furthermore, since each type of processor has its own machine code, a program written for one type of processor is not executable by any other.
1.2.1 High level languages
Today, most programs are written in high level languages, which resemble English and are therefore easier to use than machine code, but which have a limited, specialised vocabulary and a simple syntax free from ambiguity.
FORTRAN (FORmula TRANslation), introduced in 1956, was the first high level language. It has since been revised several times. Fortran 77, though not the latest version, is widely available and is compatible with later versions.
Note: although compatible with Fortran 90 standard, mind not to use old "obsolescent features" which will be progressively eliminated from future standards (see "obsolescent features" in Fortran 90/95 manuals).
High level language instructions are not executable. Instead, a high level language source program is read as input by a program called a compiler, which checks its syntax and, if it is free from errors, compiles an equivalent machine code object program. (If the source program contains syntax errors, the compiler outputs a number of messages indicating the nature of the errors and where they occur.)
Although it is in machine code, the object program is incomplete because it includes references to subprograms which it requires for such common tasks as reading input, producing output and computing mathematical functions. These subprograms are grouped together in libraries which are available for use by all object programs. To create an executable program, the object program must be linked to the subprogram libraries it requires. The executable program may then be loaded into memory and run. The steps required to compile, link and run a Fortran program are illustrated by Figure 3.
| First page |
-2- Constants and variables
- 2.1 The CHARACTER type
- 2.2 The INTEGER type
- 2.3 The REAL type
- 2.4 Variables
- 2.4.1 Explicit typing
- 2.4.2 Implicit (or default) typing
- 2.4.3 Assigning a value
- 2.4.3.1 The READ statement
- 2.4.3.2 The assignment statement
- 2.4.3.3 Type rules
Data are constants provided to a program for use in computation (processing).
Results are constants produced as a result of computation.
We have seen that all information is represented in the computer
in binary form. The type of information determines the
way in which it is represented and the operations which may be
performed on it.
2.1 The CHARACTER type
A constant of type CHARACTER, (often called a string) is a sequence of characters which may be upper case alphabetic, numeric, blanks, and the following:
+ - * / = ( ) , . ' $ :
When included in a Fortran statement, a string must be delimited by single quotes ('). A single quote may be included in a string by writing two consecutively. Only one is retained.
Example: 'WE''RE A'' JOCK TAMSON''S BAIRNS.'
2.2 The INTEGER type
Constants of type INTEGER are integer numbers. An INTEGER constant is written as a sequence of decimal digits, optionally preceded by a sign (unary + or -).
Examples:
123 +1 0 4356 -4
INTEGER constants are represented in exact form. Their
magnitude has a limit which depends on the word length of the
computer.
2.3 The REAL type
Constants of type REAL are numbers which may include a fractional part. A REAL constant is written in one of the following forms:
- An integer part written as an INTEGER constant defined as
above, followed by a decimal point, followed by a fractional part
written as a sequence of decimal digits. Either the integer or
the fractional part, but not both, may be omitted.
- An INTEGER constant or a REAL constant, followed by a decimal exponent written as the letter 'E' followed by an INTEGER constant. The constant is a power of 10 by which the preceding part is multiplied.
Examples:
+123.4 -123.4 .6E-3 (0.6x10-3) 4.6E3 (4.6x103) 7E-3 2.REAL constants are represented in approximate form. Their magnitude has a limit which depends on the word length of the computer.
Note: the explicit type length specification in bytes (8 bits) is often
specified after an "*". Examples: REAL*8 or INTEGER*4
It was an extension to Fortran 77, now replaced by the KIND
type parameter in Fortran 90.
2.4 Variables
A variable is a unique name which a Fortran program applies to a word of memory and uses to refer to it. A variable consists of one to six upper case alphabetic characters and decimal digits, beginning with an alphabetic character.
Examples:
VOL TEMP A2 COLUMN IBM370
Spaces are ignored by Fortran 77, e.g. 'COL UMN' is equivalent to 'COLUMN'.
Note: spaces becomes significant in the "free format" of Fortran 90 !
Clarity can be improved by choosing variables which suggest their usage, e.g.
DEGC MEAN STDDEVThe value of a variable is the constant stored in the word to which it refers. Each variable has a type, which stipulates the type of value it may have. The type of a variable may be specified explicitly, or assigned implicitly (by default).
The type of a variable may be assigned explicitly by a type specification statement. This has the form:
type variable_list
where type is the name of a type
and variable_list is a single variable or a list of variables, separated by commas.
The statement assigns the given type to all the variables in the list.
Examples:
INTEGER WIDTH
REAL NUM, K
Type specification statements are not compiled into executable
machine code instructions. Instead the compiler records the names
and types of the variables and reserves storage for them. Such
non-executable statements must be placed at the beginning
of a program, before the first executable statement.
2.4.2 Implicit (or default) typing
If a variable is used without being included in a type specification, its type is assigned implicitly (by default) according to the following rule:
If the variable begins with a character from I to N, its type is INTEGER. Otherwise, it is REAL.
Thus TEMP is a REAL variable, while ITEMP is an INTEGER.
Note: because a variable can be used without first being declared in a type specification, a misspelled variable is not in general detected as an error by the compiler. The program may compile and run, but produce incorrect results. Care should therefore be taken to get variable names right, and if unexpected results are obtained, variable names are one of the first things to check.
Note: it is strongly recommended to force explicit typing by
placing an IMPLICIT NONE statement before type
specifications. It was an extension to the Fortran 77 standard, but
it is now integral part of Fortran 90 standard.
2.4.3 Assigning a value
Before a variable can be used in computation, it must be assigned an initial value. This may be done by reading a value from input or by using an assignment statement.
The READ statement
The READ statement is used to assign values to variables by reading data from input. The simplest form of the READ statement is:
READ *, variable_list
where variable_list is a single variable or a list of variables separated by commas. (The asterisk will be explained later).
This statement reads constants from the terminal, separated by spaces, commas, or new lines, and assigns them in sequence to the variables in the list. Execution of the program pauses until the right number of constants has been entered.
Example:
READ *, VAR1, VAR2, VAR3
waits for three constants to be entered and assigns them in sequence
to the variables VAR1, VAR2 and VAR3.
The assignment statement
The simplest form of assignment statement is:
variable = constant
This means that the constant is assigned as a value to the variable on the left-hand-side. Note that the '=' sign has a different meaning than in algebra. It does not indicate equality, but is an assignment operator.
Examples:
TEMP = 74.5
ITEMP = 100
Type rules
Whichever method is used to assign a value to a variable, the type of the value must be consistent with that of the variable. The rules are:
- A CHARACTER value cannot be assigned to a numeric variable or vice versa.
- An INTEGER value can be assigned to a REAL variable.
The value assigned is the REAL equivalent of the integer.
Example: X = 5 is equivalent to X = 5.0 - A REAL value can be assigned to an INTEGER variable.
The value assigned is truncated by discarding the fractional
part.:
Examples:Value Assigned N = 0.9999 0 M = -1.9999 -1
| First page |
-3- Arithmetic expressions and assignment
An arithmetic expression is one which is evaluated by performing a sequence of arithmetic operations to obtain a numeric value, which replaces the expression. Arithmetic operations are denoted by the following arithmetic operators
Operator Operation + Addition or unary + - Subtraction or unary - * Multiplication / Division ** Exponentiation
Figure 4 : Arithmetic operators
An operand of an arithmetic expression may be:
- a numeric constant
- a numeric variable, which may be preceded by a unary + or -.
- an arithmetic expression in parentheses, i.e. (arithmetic_expression)
An arithmetic expression has the form:
operand [operator operand] ...
The square brackets indicate that the items operator operand are optional and the ellipsis (...) that they may be repeated indefinitely.
Spaces may be used to improve readability, but are ignored by the Fortran 77 compiler.
Examples:
3.14159
K
(A+B)*(C+D)
-1.0/X + Y/Z**2
2.0 * 3.14159*RADIUS
Restrictions:
- Two operators cannot be written consecutively.
Example: A*-B is illegal. The second factor must be made into an expression by using parentheses, viz: A*(-B). - The operands must be such that the operations are mathematically defined, e.g. dividing by zero, raising zero to a negative or zero power, or raising a negative number to a real power are all illegal.
3.1.1 Evaluation (precedence order)
An arithmetic expression is evaluated by replacing variable operands and parenthesised expressions by their values, and performing the operations indicated in sequence, using the result of each operation as an operand of the next. Clearly, the sequence in which the operations are performed is significant, as shown by the example:
4.0+6.0*2.0
which evaluates to 20.0 if the addition is performed first, or
16.0 if the multiplication is performed first.
The sequence of operations is determined by the following precedence order, in which operators on any line have equal precedence and precedence decreases downwards.
**
* /
+ - (binary and unary)
Using this precedence order, the rules for the evaluation of an arithmetic expression may be stated as follows:
- Evaluation begins with the leftmost operand.
- A variable operand, or a parenthesised expression, is evaluated before the following rules are applied.
- Exponentiation is performed from right to left.
Example: 2**3**2 evaluates to 512 (2**9). - All other operations are performed from left to right unless
the operator precedence rules stipulate the opposite.
Examples:
4.0+6.0*2.0 evaluates to 16.0
(4.0+6.0)*2.0 evaluates to 20.0
3.1.2 Type rules for arithmetic expressions
Subject to the restrictions noted under 'Restrictions:' above, REAL and INTEGER operands may be freely mixed in an arithmetic expression. The type of the expression's value is determined by applying the following rules to each operation performed in its evaluation:
- If both operands are of the same type, the resulting value is also of that type.
- If one operand is INTEGER and the other REAL,
the INTEGER operand is converted to its REAL
equivalent before the operation is performed, and the resulting
value is REAL.
This rule is inconsistent with the rules of arithmetic, in which dividing one integer by another (e.g. 7/5) or raising an integer to an integer power (e.g. 2-1) does not always result in an integer. Fortran deals with such cases by truncating, as described under "Type rules" (in the above paragraph "Assigning a value"), to obtain an INTEGER value.
Examples:
Value 99/100 0 7/3 2 -7/3 -2 N**(-1) 0 N**(1/2) 1 100*9/5 180 9/5*100 100The last two examples show that the ordering of * and / operators with INTEGER operands is significant. It is usually best to avoid dividing one integer by another unless there are special reasons for doing so.
As noted at the beginning of the chapter, the value of an arithmetic expression is assigned to a numeric variable in a statement of the form:
numeric_variable = arithmetic_expression
If the type of the expression differs from that of the variable, the rules listed under 'Type rules' (in the above paragraph "Assigning a value"), are applied, i.e.
Expression Variable Rule type type INTEGER REAL Convert to REAL REAL INTEGER Truncate
| First page |
-4- A Simple Program
- 4.1 The PRINT statement
- 4.2 The PROGRAM statement
- 4.3 The END and STOP statements
- 4.4 Program layout (fixed format)
- 4.5 Comments
- 4.6 A simple program
We are now ready to write a simple Fortran program. All that is required is some information on printing output, program layout and a few simple statements.
Output can be printed using the PRINT statement, which is very similar to the READ statement shown above:
PRINT *, output_list
where output_list is a single constant, variable or expression or a list of such items, separated by commas.
Example: PRINT *,'THE RESULTS ARE', X ,'AND',Y
The PRINT statement prints the output list on the terminal screen in a standard format. Later, we shall consider more flexible output statements which give us greater control over the appearance of the output and the device where it is printed.
A program can optionally be given a name by beginning it with a single PROGRAM statement. This has the form:
PROGRAM program_name
where program_name is a name conforming to the rules for Fortran variables.
4.3 The END and
STOP
statements
Each program must conclude with the statement END, which marks the end of the program. There must be no previous END statement.
The statement STOP stops execution of the program. In Fortran 77/90, but not in previous versions, END also has this effect. Therefore, if execution is simply to stop at the end of the program, STOP is optional. However, one or more STOP statements may be written earlier, to stop execution conditionally at points other than the end.
4.4 Program layout (fixed format of Fortran 77)
When Fortran was introduced, punched cards were a common input medium. Fortran was designed to make use of the cards' 80-column layout by ignoring spaces and reserving different fields of the card for different purposes. Although cards are no longer used, Fortran still uses this column-based layout.
All Fortran statements must be written in columns 7 to 72. A statement ends with the last character on the line, unless the next line has any character other than 0 in column 6. Any such character indicates that columns 7 to 72 are a continuation of the previous line.
Columns 73 to 80 are ignored by the compiler. Originally, these columns were used to print sequence numbers on the cards, but now they are normally unused.
Columns 1 to 5 are reserved for statement labels. These are optional unique unsigned non-zero integers used to provide a reference to statements.
The layout rules are summarised in Figure 5.
Columns Usage
1-5 Statement labels
6 Continuation character or blank
7-72 Fortran statements
73-80 Unused
Figure 5: Fortran layout
Note: the fixed format source of Fortran 77 is an obsolescent
feature of Fortran 95. It must be repaced by the new free format.
4.5 Comments
The letter 'C' or an asterisk in column one causes the compiler to ignore the rest of the line, which may therefore be used as a comment to provide information for anyone reading the program.
The following example uses the Fortran statements introduced so far to solve a simple problem.
Example 1: a driver fills his tank with petrol before setting out on a journey. Each time he stops for petrol he puts in 40 litres. At his destination, he fills the tank again and notes the distance he has travelled in kilometres. Write a program which reads the distance travelled, the number of stops and the amount of petrol put in at the end of the journey, and prints the average petrol consumption in kilometres per litre, rounded to the nearest litre.
1| PROGRAM PETROL 2| INTEGER STOPS, FILLUP 3| C 4| C THESE VARIABLES WOULD OTHERWISE BE TYPED REAL BY DEFAULT 5| C ANY TYPE SPECIFICATIONS MUST PRECEDE THE FIRST EXECUTABLE STATEMENT 6| C 7| READ *, KM,STOPS,FILLUP 8| USED = 40*STOPS + FILLUP 9| C COMPUTES THE PETROL USED AND CONVERTS IT TO REAL 10| KPL = KM/USED + 0.5 11| C 0.5 IS ADDED TO ENSURE THAT THE RESULT IS ROUNDED 12| PRINT *, 'AVERAGE KPL WAS',KPL 13| ENDFigure 6: Petrol consumption program
This program illustrates some of the points about type conversion made in the previous chapter. In line 8, the number of litres of petrol used is computed. The computed value is of type INTEGER, but is converted to REAL when assigned to the REAL variable USED.
In line 10, the expression KM/USED is evaluated as REAL, but would be truncated, not rounded, when assigned to the INTEGER variable KPL. Adding 0.5 before truncating has the effect of rounding up or down. This is a useful rounding method. It is illustrated further below.
KM/USED KM/USED + 0.5 KPL
12.0 12.5 12 12.4 12.9 12 12.5 13.0 13 12.9 13.4 13
| First page |
-5- Control structures - Conditional execution
- 5.1 The LOGICAL type
- 5.2 Logical expressions
- 5.2.1 Relational expressions
- 5.2.2 Composite logical expressions
- 5.3 Logical assignment
- 5.4 The logical IF statement
- 5.5 The block IF structure
The Fortran statements covered so far are enough to allow us to read information, evaluate arithmetic expressions and print results. It is hardly necessary to write a program to perform such tasks, which can usually be more easily done using a calculator.
The main advantages of a computer are its ability to:
- execute alternative sequences of instructions depending on a condition (conditional execution).
- execute a sequence of instructions repeatedly while or until a condition is satisfied (iteration).
This chapter deals with conditional execution while iteration is covered in Chapter 6.
The need for conditional execution is illustrated by the following problem:
Example 1: write a program to read the coefficients of a quadratic equation and print its roots.
Solution: the roots of the quadratic equation
ax² + bx +c
are given by the formula
The outline of the program is:
- Read the coefficients a, b and c
- Evaluate b² - 4ac
- If b² - 4ac exceeds zero then
- Compute and print two distinct real roots.
- Compute and print two coincident real roots.
- Print message: 'No real roots'.
In step 3, the program must test conditions such as "b² -4ac exceeds zero".
To express such conditions, Fortran uses another type, the LOGICAL type.
There are two LOGICAL constants, defined as .TRUE. and .FALSE..
A LOGICAL variable can be assigned either of these values. It may not be assigned a value of any other type. Each LOGICAL variable must be declared in a LOGICAL type specification statement, which must occur, like all other type specifications, before the first executable statement.
Example: the LOGICAL variable ERROR could be declared and initialised by the statements:
LOGICAL ERROR
ERROR = .FALSE.
A logical expresssion is one which evaluates to one of the LOGICAL constants .TRUE. or .FALSE.. Thus the simplest logical expressions are the LOGICAL constants themselves, and LOGICAL variables.
5.2.1 Relational expressions
A relational expression is a logical expression which states a relationship between two expressions, evaluating to .TRUE. if the relationship applies or .FALSE. otherwise. For the present, we shall consider only relationships between arithmetic expressions. (As we shall see later, Fortran can also deal with relationships between CHARACTER expressions.)
A relational expression has the form:
arithmetic_expression relational_operator arithmetic_expression
The relational operators are:
Meaning
.LT. Less than
.LE. Less than or equal to
.EQ. Equal to
.NE. Not equal to
.GE. Greater than or equal to
.GT. Greater than
Thus examples of relational expressions are:
N.GE.0
X.LT.Y
B**2 - 4*A*C .GT. 0.
Notes:
- Relational operators have lower precedence than arithmetic operators. Therefore, in evaluating a relational expression, the arithmetic expressions are evaluated before the comparison indicated by the relational operator is made.
- The two arithmetic expressions may be of different type (i.e. one INTEGER and one REAL). In this case, the INTEGER expression is converted to REAL form before the comparison is made.
5.2.2 Composite logical expressions
It is often necessary to express a condition which combines two or more logical expressions. For example, to check that the value of a variable lies within a given range, we should have to check that it is greater than the lower limit AND less than the upper limit. Such conditions are expressed in Fortran by composite logical expressions, which have the form:
L1 logical_operator L2
where L1 and L2 are logical expressions (relational or composite). The logical operators and their meanings are shown below. The second column indicates the conditions under which a composite logical expression as above evaluates to .TRUE..
Meaning
.AND. Both L1 and L2 are .TRUE.
.OR. Either L1 or L2 or both are .TRUE.
.EQV. Both L1 and L2 have the same value (.TRUE. or .FALSE.)
.NEQV. L1 and L2 have different values (one .TRUE. and one .FALSE.)
Thus the following composite logical expression would evaluate to .TRUE. if the value of the variable X lay within a range with non-inclusive limits MIN and MAX:
X.GT.MIN .AND. X.LT.MAX
There is one further logical operator .NOT., which unlike the others, takes only one operand, which it precedes. The expression .NOT.L is .TRUE. if the logical expression L is .FALSE. and vice versa.
As with arithmetic operators, precedence rules are required to define the interpretation of expressions like:
.NOT. L1 .OR. L2
which could evaluate to .TRUE. under either of the following conditions,
depending on the order of evaluation:
- L1 is .FALSE. or L2 is .TRUE.
- L1 and L2 are both .FALSE.
The precedence order is shown by the following list, in which precedence decreases downwards.
| Arithmetic operators |
| Relational operators |
| .NOT. |
| .AND. |
| .OR. |
| .EQV. and .NEQV. |
Thus (1) is the correct interpretation of the above expression.
As in arithmetic expressions, parentheses can be used to group partial logical expressions and change the order of evaluation. Thus
.NOT.(L1.OR.L2)
would be evaluated according to interpretation (2).
Parentheses can also be used to improve clarity, even when not logically required, e.g.
(A.LT.B) .OR. (C.LT.D)
5.3 Logical assignment
The value of a logical expression can be assigned to a variable of type LOGICAL, e.g.
LOGICAL VALID
VALID = X.GT.MIN .AND. X.LT.MAX
Logical expressions are more commonly used in logical IF statements and structures.
5.4 The logical IF statement
The logical IF statement is used to execute an instruction conditionally. It has the form:
IF (logical_expression) executable_statement
where executable_statement is an executable Fortran statement other than another IF statement or a DO statement (see Chapter 6).
The statement is executed by evaluating logical_expression and executing executable_statement if it evaluates to .TRUE..
Example: IF (A.LT.B) SUM = SUM + A
5.5 The block IF structure
The logical IF statement is of limited usefulness, as it permits only the execution of a single instruction depending on a single condition. The block IF structure is more powerful, permitting the conditional execution of one of a number of alternative sequences of instructions. It may be described informally as:
- an IF block, followed by:
- one or more optional ELSE IF blocks, followed by:
- an optional ELSE block, followed by:
- END IF
More formally, the structure is:
IF (Lo) THEN So ELSE IF (Li) THEN Si ..... ... ELSE Sn END IFwhere:
- Lo and Li are logical expressions.
- So, Si and Sn are sequences of Fortran statements.
- The ELSE IF and ELSE blocks are optional and the ellipsis (...) indicates that it may be repeated indefinitely.
The structure is executed as follows:
Lo is evaluated. If it evaluates to .TRUE., the sequence So is executed and execution continues with the statement following END IF.
Otherwise:
If there are any ELSE IF clauses, each Li is evaluated, until either:
- An Li evaluates to .TRUE. The sequence Si
is executed and execution continues with the statement following
END IF.
or - The last Li evaluates to .FALSE. Execution continues.
- If there is an ELSE clause, the sequence Sn is executed.
- Execution continues with the statement following END IF.
Thus, a simple block IF structure is:
IF (A.LT.B) THEN
SUM = SUM + A
PRINT *, SUM
END IF
which is equivalent to the IF statement shown earlier.
A more realistic example is the following:
Example: an employee is paid at the standard rate for the first 40 hours of work, at time and a half for the next 10, and at double time for any hours in excess of 50. If the variable HRS represents the hours worked and RATE the standard rate then the employee's salary is computed by the block IF structure:
IF (HRS.LE.40) THEN
SALARY = HRS*RATE
ELSE IF (HRS.LE.50) THEN
SALARY = 40.0*RATE + (HRS-40.0)*RATE*1.5
ELSE
SALARY = 40.0*RATE + 10.0*RATE*1.5 + (HRS-50.0)*RATE*2.0
END IF
Note the use of indentation to clarify the structure.
We are now in a position to complete the quadratic roots program of Example 1, but first the outline should be altered as follows:
- Because REAL values are approximations, exact comparisons
involving them are unreliable. The relational expressions in our
program should therefore be reformulated using a variable
e (for error) to which a small positive value
(1.0E-9) has previously been assigned.
The expression "b² - 4ac exceeds zero" in step 3 should be replaced by:
b² - 4ac > eSimilarly, the expression "b² - 4ac is equal to zero" in step 3 should be replaced by:
-e <= b² -4 ac <= eHowever, the expression is evaluated only if b² -4 ac > e has previously been evaluated as false, which of course implies b² -4 ac <= e. Therefore, all that is required is:
-e <= b² -4 acComparisons involving REAL values should always be expressed in this way.
- The expression for the roots includes a divisor of 2a. Therefore if a has a value of zero, the evaluation of this expression will cause the program to fail with an arithmetic error. The program should prevent this by testing and printing a suitable message if it is zero. Again, the test should be expressed using the error variable e.
In general, programs should be designed to be robust, i.e. they should take account of any exceptional data values which may cause the program to fail, and take steps to prevent this.
The program outline now becomes:
- Assign a small positive value to e.
- Read the coefficients a, b and c.
- If -e <= a <= e then
- Print message: 'First coefficient must be non-zero'.
- Otherwise:
- Evaluate b² - 4ac
- If b² -4 ac > e then
- Compute and print two distinct real roots.
- Compute and print two coincident real roots.
- Print message: 'No real roots.'
Now that the outline is complete, the program can be easily written:
PROGRAM QUAD
E = 1E-9
READ *, A,B,C
IF (A.GE. -E .AND. A.LE.E) THEN
PRINT *, 'FIRST COEFFICIENT MUST BE NON-ZERO.'
ELSE
S = B**2 - 4*A*C
IF (S.GT.E) THEN
D = S**0.5
X1 = (-B+D)/(2*A)
X2 = (-B-D)/(2*A)
PRINT *, 'TWO DISTINCT ROOTS:' X1 'AND' X2
ELSE IF (S.GT. -E) THEN
X = -B/(2*A)
PRINT *, 'TWO COINCIDENT ROOTS',X
ELSE
PRINT *, 'NO REAL ROOTS.'
END IF
END IF
END
Figure 7: Quadratic roots program
Note that most of the program consists of a block IF structure, with a second block IF included in its ELSE clause. The embedding of one structure within another in this way is called nesting.
Once again, indentation has been used to clarify the structure.
| First page |
-6- Control structures - Iteration
- 6.1 The GOTO statement
- 6.2 Count controlled loops
- 6.3 The DO-loop
- 6.3.1 Execution
- 6.3.2 Restrictions and other notes
- 6.3.3 The control variable
- 6.3.4 Nested DO loops
The last chapter showed how a sequence of instructions can be executed once, if a condition is true. The need also frequently arises to execute a sequence of instructions repeatedly, while a condition is true, or until a condition becomes true. Such repetitive execution is called iteration.
Write a program to read the marks (notes in french) of a class of students in an exam, print the number of marks and compute and print the average mark. The marks are to be read one at a time, with a 'dummy' mark of 999 marking the end.
The outline of the program is:
- Initialise the total mark to zero.
- Initialise a count of the number of marks to zero.
- Read the first mark.
- While the current mark is not 999, repeat:
- Increment the count.
- Add the mark to the total.
- Read the next mark.
- If the count exceeds zero then
- Print the count.
- Compute and print the average mark.
- Otherwise:
- Print message: 'No marks read.'
Unlike Fortran 90 and other more modern programming languages,
Fortran 77 lacks a while
structure as such, but the effect can be obtained using an IF
structure and a new statement, the GOTO.
6.1 The GOTO statement
The GOTO statement has the form:
GOTO label
where label is the label of an executable statement, with certain restrictions which will be considered later.
A GOTO statement causes the flow of execution to 'jump' to the labelled statement and resume from there.
We can use a GOTO to complete the program of Example 1:
PROGRAM AVMARK
INTEGER TOTAL,COUNT
TOTAL = 0
COUNT = 0
READ *, MARK
10 IF (MARK.NE.999) THEN
COUNT = COUNT+1
TOTAL = TOTAL+MARK
READ *, MARK
GOTO 10
END IF
IF (COUNT.GT.0) THEN
AVER = 1.0*TOTAL/COUNT
C MULTIPLY BY 1.0 TO CONVERT TO REAL AND AVOID TRUNCATION
PRINT *, COUNT, 'MARKS WERE READ.'
PRINT *, 'AVERAGE MARK IS', AVER
ELSE
PRINT *, 'NO MARKS WERE READ.'
END IF
END
Figure 8: Average mark program
Exercise:
the average mark program provides for the possibility that the
data consists only of the terminator 999, but does no other checking.
If the range of valid marks is 0 to 100, alter the program to
check the validity of each mark, printing a suitable message if
it is invalid, and print a count of any invalid marks with the
results.
6.2 Count controlled loops
Any iterative structure is usually called a loop. As in example 1, a loop of any kind can be constructed using an IF structure and one or more GOTO's.
Often a loop is controlled by a variable which is incremented or decremented on each iteration until a limiting value is reached, so that the number of iterations is predetermined. Such a loop is shown in Example 2.
Example 2: write a program to print a table of angles in degrees and their equivalents in radians, from 0 to 360 degrees, in steps of 10 degrees.
The program outline is:
- Initialise the angle in degrees to 0.
- While the angle does not exceed 360 degrees, repeat:
- Compute the equivalent in radians.
- Print the angle in degrees and radians.
- Increment the angle by 10 degrees.
and the program is:
PROGRAM CONVRT
INTEGER DEGREE
CONFAC = 3.141593/180.0
C CONVERSION FACTOR FROM DEGREES TO RADIANS
DEGREE = 0
10 IF (DEGREE .LE. 360) THEN
RADIAN = DEGREE*CONFAC
PRINT *, DEGREE,RADIAN
DEGREE = DEGREE + 10
GOTO 10
END IF
END
Figure 9: Degrees to radians conversion program (version 1)
Loops of this kind occur frequently. Their essential features are:
- A loop control variable (DEGREE in the above example) is assigned an initial value before the first iteration.
- On each iteration, the control variable is incremented or decremented by a constant.
- Iteration stops when the value of the control variable passes a predefined limit.
Fortran provides for such loops with a structure called a DO loop, which is more concise and readable than a construction using IF and GOTO.
6.3 The DO-loop
A DO loop is a sequence of statements beginning with a DO statement. This has the form:
DO label, var = e1, e2, [,e3]
the square brackets indicating that ',e3' may be omitted.
label is the label of an executable statement sequentially following the DO statement called the terminal statement of the DO loop.
var is an INTEGER or REAL (obsolete Fortran 90 feature) variable called the loop control variable.
e1, e2 and e3 are arithmetic expressions (i.e. INTEGER or REAL constants, variables or more complex expressions).
The sequence of statements beginning with the statement immediately following the DO statement and ending with the terminal statement is called the range of the DO loop.
A DO loop is executed as follows:
- The expressions e1, e2 and e3 are evaluated
and if necessary converted to the type of var. If e3
is omitted, a value of 1 is used. The resulting values are called
the parameters of the loop. We shall call them initial,
limit and increment respectively.
- initial is assigned as a value to var.
- var is compared with limit, the test depending
on the value of increment as follows:
Condition tested
increment > 0 ---> var <= limit
increment < 0 ---> var >= limitIf the condition tested is "true", then:
- the range of the DO loop is executed,
- var is incremented by increment,
- control returns to step 3.
Examples:
DO 10, I = 1,5
causes the range of statements beginning with the next and ending with the statement labelled 10 to be executed 5 times.
DO 10, I = 0,100,5
causes the range to be executed 21 times for values of I of 0,5,10...100.
DO 10, I = 100,0,-5
causes the range to be executed 21 times for values of I of 100,95...0.
DO 10, I = 0,100,-5
In this case, the range is not executed at all, as the test in step 3 fails for the initial value of I.
DO 10, J = I,4*N**2-1,K
Here, e1, e2 and e3 are more complex expressions.
We can now rewrite the program of Example 2 using a DO loop. The outline becomes:
- Repeat for angles in degrees from 0 to 360 in steps of 10:
- Compute the equivalent in radians.
- Print the angle in degrees and radians.
and the program follows:
PROGRAM CONVRT
INTEGER DEGREE
CONFAC = 3.141593/180.0
C CONVERSION FACTOR FROM DEGREES TO RADIANS
DO 10, DEGREE = 0,360,10
RADIAN = DEGREE*CONFAC
PRINT *, DEGREE,RADIAN
10 CONTINUE
END
Figure 10: Degrees to radians conversion program (version 2)
This is clearer and more concise than version 1. Note the use of indentation to clarify the loop structure.
6.3.2 Restrictions and other notes
To protect the integrity of the loop structure, there are various restrictions affecting DO loops.
- Increment must not be zero.
- The terminal statement must be one which is self-contained and
allows execution to continue at the next statement. This rules
out STOP, END and another DO statement.
It is often convenient to end a DO loop with a
CONTINUE statement, which has no
effect whatever, serving only to mark the end of the loop.
- The range of a DO loop can be entered only via the initial DO
statement. Thus a GOTO cannot cause a jump into the range of a
DO loop. However, GOTOs can be included in the range
to jump to statements either inside or outside it. In the latter case,
this can cause iteration to stop before the control variable reaches
the limiting value.
Examples:
GOTO 10 . . . DO 20, I = 1,5 . . . 10 . . . . . . 20 CONTINUEis wrong, but
DO 20, I = 1,5 . . . 10 . . . . . . IF (...) GOTO 10 . IF (...) GOTO 30 . . . 20 CONTINUE . . . 30 . . .is all right.
- The control variable can be freely used in expressions in the
range of the loop (as in Figure 10) but it cannot be assigned
a value.
- The loop parameters are the values of the expressions e1,
e2 and e3 on entry to the loop. The expressions themselves
are not used. Therefore if any of e1, e2 and e3
are variables, they can be assigned values within the loop without
disrupting its execution.
As explained under 'Execution' the control variable is incremented and tested at the end of each iteration. Thus, unless iteration is interrupted by a GOTO, the value of the control variable after execution of the loop will be the value which it was assigned at the end of the final iteration. For example, in a loop controlled by the statement:
DO 10, I = 0,100,5the control variable I is incremented to exactly 100 at the end of the 20th iteration. This does not exceed limit, so another iteration is performed. I is then incremented to 105 and iteration stops, with I retaining this value.
If the control variable is REAL, inconsistent results may be obtained unless allowance is made for approximation. For example, in a loop controlled by:
DO 10, C = 0,100,5the control variable C is incremented at the end of the 20th iteration to a value of approximately 100. If it is less, execution continues for a further iteration, but if it is greater, iteration stops.
To avoid such effects, a higher value of limit should be used, e.g.
DO 10, C = 0,101,5Note: REAL control variable is a Fortran 90 obsolescent feature.
6.3.4 Nested DO loops
DO loops, like IF structures, can be nested, provided that there is no overlapping. (i.e. that the range of each nested loop is entirely within the range of any loop in which it is nested).
Example:
Valid Invalid DO 20 ... DO 20 ... ... ... DO 10 ... DO 10 ... ... ... 10 CONTINUE 20 CONTINUE ... ... 20 CONTINUE 10 CONTINUEThe following provides a simple, if not very useful example of a nested loop structure.
Example 3:
Write a program to print a set of multiplication tables from 2 times up to 12 times.
The outline is:
Repeat for I increasing from 2 to 12:
- Print table heading.
- Repeat for J increasing from 1 to 12:
- Print I 'times' J 'is' I*J
and the program is:
PROGRAM TABLES DO 20, I = 2,12 PRINT *,I,' TIMES TABLE' DO 10, J = 1,12 10 PRINT *,I,' TIMES',J,' IS',I*J 20 CONTINUE ENDFigure 11: Multiplication tables programThere is no logical need for the CONTINUE statement in this program as nested loops can share a common terminal statement. Thus the program could be rewritten as:
PROGRAM TABLES DO 10, I = 2,12 PRINT *,I,' TIMES TABLE' DO 10, J = 1,12 10 PRINT *,I,' TIMES',J,' IS',I*J ENDHowever, to clarify the structure, it is better to use separate terminal statements and indentation as in the first version.
Note: sharing terminal statement or eliminating the terminal CONTINUE is an obsolescent feature of Fortran 90.
First page
-7- Arrays
- 7.1 Array declarations
- 7.2 Use of arrays and array elements
- 7.3 Initialising an array
- 7.3.1 The DATA statement
- 7.3.2 The implied DO list
- 7.4 Input and output of arrays
- 7.5 Multi-dimensional arrays
All our programs so far have required the storage of only a few values, and could therefore be written using only a few variables. For example, the average mark program of Figure 8 required only variables for a mark, the total mark, the count of the marks and the average. When large numbers of values have to be stored, it becomes impracticaI or impossible to use different variables for them all. If the average mark program were rewritten to compute average marks for five subjects, we should require five variables, say MARK1 ... MARK5 for the marks, five variables for the totals, and five for the averages. This could be done, but the program would be rather repetitive. The situation is even worse if, after computing the averages, the program is required to print a list showing, for each student and subject, the student's mark and the difference between the mark and the average. This could conceivably be done if the number of students were given in advance, but the program would be extremely cumbersome. If, as in the example, the number of students is not given but determined by counting, the task is impossible, as there is no way of knowing how many variables will be required.
We need to store all the marks in order in a list or other structure to which we can apply a name, and refer to individual marks by a combination of the name and a number or numbers indicating the position of a mark in the list or structure.
In mathematics, an ordered list of n items is called a vector of dimension n. If the vector is denoted by v, the items, usually called the components or elements of the vector, are denoted by v
1 , v 2 , v 3 , ..., v n . Fortran uses a structure similar to a vector called an array. An array A of dimension N is an ordered list of N variables of a given type, called the elements of the array. In Fortran, the subscript notation used for the components of a vector is not available. Instead the elements are denoted by the name of the array followed by an integer expression in parentheses. Thus, the elements of A are denoted by A(1), A(2),... A(N). The parenthesised expressions are called array subscripts even though not written as such.
A subscript can be any arithmetic expression which evaluates to an integer. Thus, if A, B, and C are arrays, the following are valid ways of writing an array element:
A(10) B(I+4) C(3*I+K)7.1 Array declarations
Since an array is a list of variables, it obviously requires several words or other units of storage. Each array must therefore be declared in a statement which tells the compiler how many units to reserve for it. This can be done by including the array name in a type specification followed by its dimension in parentheses. For example:
INTEGER AGE(100),NUM(25),DEGThis reserves 100 words of storage for array AGE, 25 for array NUM, and one word for the variable DEG. All three items are of type INTEGER.
Space can also be reserved for arrays by the DIMENSION statement, which reserves storage using a similar syntax, but includes no information about type. Thus, if this method is used, the type is either determined by the initial letter of the array or assigned by a separate type specification. Therefore, the equivalent to the above using a DIMENSION statement is:
INTEGER AGE,DEG DIMENSION AGE(100),NUM(25)(NUM is typed as INTEGER by default).
DIMENSION statements, like type specifications, are non-executable and must be placed before the first executable statement.
When this form of declaration is used in a type or DIMENSION statement the upper and lower bounds for the subscript are 1 and the dimension respectively. Thus, AGE in the above example may have any subscript from 1 to 100. Arrays can also be declared to have subscripts with a lower bound other than 1 by using a second form of declaration in which the lower and upper bounds are given, separated by a colon. For example:
REAL C(0:20) INTEGER ERROR(-10:10)reserves 21 words of storage for each of the arrays C and ERROR and stipulates that the subscripts of C range from 0 to 20 inclusive, while those of ERROR range from -10 to 10.
Although the declaration stipulates bounds for the subscript, not all compilers check that a subscript actually lies within the bounds. For example, if NUM is declared as above to have a subscript from 1 to 25, a reference to NUM(30)may not cause an error. The compiler may simply use the 30th word of storage starting from the address of NUM(1) even though this is outside the bounds of the array. This can cause unpredictable results. Care should therefore be taken to make sure that your subscripts are within their bounds.
7.2 Use of arrays and array elements
Array elements can be used in the same way as variables, their advantage being that different elements of an array can be referenced by using a variable as a subscript and altering its value, for example by making it the control variable of a DO loop. This is illustrated in the following sections.
The array name without a subscript refers to the entire array and can be used only in a number of specific ways.
7.3 Initialising an array
Values can be assigned to the elements of an array by assignment statements, e.g.
NUM(1) = 0 NUM(2) = 5If all the elements are to have equal values, or if their values form a regular sequence, a DO loop can be used. Thus, if NUM and DIST are arrays of dimension 5:DO 10, I = 1,5 NUM(I) = 0 10 CONTINUEinitialises all the elements of NUM to 0, while:DO 10, I = 1,5 DIST(I) = 1.5*I 10 CONTINUEassigns the values 1.5, 3.0, 4.5, 6.0 and 7.5 to DIST(1),DIST(2),DIST(3),DIST(4) and DIST(5) respectively.7.3.1 The DATA statement
The DATA statement is a non-executable statement used to initialise variables. It is particularly useful for initialising arrays. It has the form:
DATA variable_list/constant_list/ [,variable_list/constant_list/] ...
(The square brackets and ellipsis have their usual meaning.)
Each variable_list is a list of variables, and each constant_list a list of constants, separated by commas in each case. Each constant_list must contain the same number of items as the preceding variable_list and corresponding items in sequence in the two lists must be of the same type.
The DATA statement assigns to each variable in each variable_list a value equal to the corresponding constant in the corresponding constant_list. For example:
DATA A,B,N/1.0,2.0,17/assigns the values 1. and 2. respectively to the REAL variables A and B, and 17 to the INTEGER variable N.A constant may be repeated by preceding it by the number of repetitions required (an integer) and an asterisk. Thus:
DATA N1,N2,N3,N4/4*0/assigns a value of zero to each of the variables N1,N2,N3 and N4.Items in a variable_list may be array elements. Thus, if A is an array of dimension 20, the DATA statement:
DATA A(1),A(2),A(3),A(4)/4*0.0/,A(20)/-1.0/assigns a value of zero to the first four elements, -1.0 to the last element, and leaves the remaining elements undefined.7.3.2 The implied DO list
When a large number of array elements have to be initialised, we can avoid writing them all individually by using an implied DO list.
An implied DO list is used in a DATA statement or an input/output statement to generate a list of array elements. The simplest form of implied DO list is:
(dlist, int=c1,c2[,c3])
where dlist is a list of array elements separated by commas. The expresssion: int=c1,c2[,c3] has a similar effect to the expression: var=e1,e2,[,e3] in a DO loop, but int must be a variable of type INTEGER, and c1,c2 and c3 must be constants or expressions with constant operands. The implied DO variable int is defined only in the implied DO list, and is distinct from any variable of the same name used elsewhere.
The implied DO list expands dlist by repeating the list for each value of int generated by the loop, evaluating the array subscripts each time. Thus:
DATA (A(I),I=1,4)/4*0.0/,A(20)/-1.0/has the same effect as the previous example.
A more complex use of an implied DO list is shown by the example:
DATA (A(I),A(I+1),I=1,19,3)/14*0.0/,(A(I),I=3,18,3)/6*1.0/which assigns a value of zero to A(1),A(2), A(4),A(5), ... A(19),A(20) and a value of 1.0 to every third element A(3),A(6), ... A(18) .
Finally, an entire array can be initialised by including its name, without a subscript, in variable_list in a DATA statement. This is equivalent to a list of all its elements in sequence. Thus, if A has dimension 20, all the elements of A are initialised to zero by:
DATA A/20*0.0/DATA statements can be placed anywhere in a program after any specifications. In the interests of clarity, it is probably best to put them immediately before the first executable statement. Wherever they may be, they cause initialisation when the program is loaded (before execution begins). Therefore they can only be used to initialise variables and not to re-assign values to them throughout execution of the program. For this purpose, assignment statements or READ statements must be used.Note: not placing DATA statements before the first executable statement is an obsolescent feature of Fortran 90.
7.4 Input and output of arrays
Array elements and array names can be used in input/output statements in much the same way as in DATA statements. Thus, input and output lists can include:
- array elements.
- array names (equivalent to all the elements in sequence).
- implied DO lists.
Implied DO lists in input/output statements differ in two respects from those in DATA statements:
- In output statements, dlist can include any
output list item. For example:
PRINT *, (A(I),'ABC', K, I=1,4)will print the values of A(1)...A(4) followed in each case by 'ABC' and the value of K. - The loop parameters need not be constants or constant expressions,
but can include variables (INTEGER or REAL)
provided that these have been assigned values, e.g.
N = 5 . . . PRINT *,(A(I),I=1,N)
READ *, N, (A(I),I=1,N)If variables are used in this way, care should be taken to ensure that they lie within the subscript bounds of the array, as in the following example:REAL A(20) . . . READ *, N IF (N.GE.1 .AND. N.LE.20) THEN READ *, (A(I),I=1,N) ELSE PRINT *, N, 'EXCEEDS SUBSCRIPT BOUNDS.' END IFWe can now return to the exam marks problem mentioned at the beginning of the chapter.Example 1: write a program to read the marks of a class of students in five papers, and print, for each paper, the number of students sitting it and the average mark. The marks are to be read as a list of five marks in the same order for each student, with a negative mark if the student did not sit a paper. The end of the data is indicated by a dummy mark of 999.
The outline of the program is:
- Initialise the total mark for each of the five papers and a count of the number of students sitting it.
- Read five marks for the first student.
- While the first mark is not 999, repeat:
- For each of the five marks repeat:
- If the mark is not negative then:
- Increment the count of students sitting that paper.
- Add the mark to the total for that paper.
- If the mark is not negative then:
- Read five marks for the next student.
- For each of the five marks repeat:
- Repeat for each of five papers:
- If the count of students sitting the paper exceeds zero then:
- Compute the average mark for the paper.
- Print the number of the paper, the number of students sitting it, and the average mark.
- Otherwise
- Print a message: 'No students sat paper number' paper_number
- If the count of students sitting the paper exceeds zero then:
We shall use arrays MARK, COUNT and TOTAL to store the five marks for a student, a count of students sitting each paper and the total mark for each paper respectively. The program follows.
PROGRAM EXAM INTEGER MARK(5),TOTAL(5),COUNT(5) DATA COUNT/5*0/,TOTAL/5*0/ READ *,(MARK(I),I=1,5) 10 IF (MARK(1).NE.999) THEN DO 20, I=1,5 IF (MARK(I).GE.0) THEN COUNT(I) = COUNT(I)+1 TOTAL(I) = TOTAL(I)+MARK(I) END IF 20 CONTINUE READ *,(MARK(I),I=1,5) GOTO 10 END IF DO 30, I=1,5 IF (COUNT(I).GT.0) THEN AVMARK = 1.0*TOTAL(I)/COUNT(I) C MULTIPLY BY 1.0 TO CONVERT TO REAL AND AVOID TRUNCATION PRINT *,COUNT(I),' STUDENTS SAT PAPER NUMBER',I PRINT *,'THE AVERAGE MARK WAS', AVMARK ELSE PRINT *,'NO STUDENTS SAT PAPER NUMBER',I END IF 30 CONTINUE ENDFigure 12: Exam marks programOne problem with this program is that if the last line of input consists of the single terminating value of 999, the statement: READ *,(MARK(I),I=1,5) will wait for another four values to be entered. This can be avoided by following 999 by a '/' character, which is a terminator causing the READ statement to ignore the rest of the input list.
7.5 Multi-dimensional arrays
Suppose now that the exam marks program is to be altered to print a list of all the marks in each paper, with the differences between each mark and the average for the paper. This requires that all the marks should be stored. This could be done by making the dimension of MARK large enough to contain all the marks, and reserving the first five elements for the first student's marks, the next five for the second student's marks and so on. This would be rather awkward.
The problem could be dealt with more easily if we could add a second subscript to the MARK array to represent the number of each student in sequence. Our array could then be declared either by:
INTEGER MARK(5,100)or by:INTEGER MARK(100,5)and would reserve enough space to store the marks of up to 100 students in 5 subjects.In fact, Fortran arrays can have up to seven dimensions, so the above declarations are valid. The subscript bounds are specified in the same way as for one-dimensional arrays. For example:
REAL THREED(5,0:5,-10:10)declares a three-dimensional array of type REAL, with subscript bounds of 1...5, 0...5 and -10...10 in that order.An array element must always be written with the number of subscripts indicated by the declaration.
When multi-dimensional array elements are used in an implied DO list, multiple subscripts can be dealt with by including nested implied DO lists in dlist, for example:
READ *, (A(J),(MARK(I,J),I=1,5),J=1,100)Here, dlist contains two items, A(J) and the implied DO list (MARK(I,J),I=,5) .This inner implied DO list is expanded once for each value of J in the outer implied DO list. Thus the above READ statement reads values into the elements of A and MARK in the order:A(1), MARK(1,1), MARK(2,1),... MARK(5,1) A(2), MARK(1,2), MARK(2,2),... MARK(5,2) .. .. .. .. A(100), MARK(1,100), MARK(2,100),... MARK(5,100)The unsubscripted name of a multi-dimensional array can be used, like that of a one-dimensional array, in input/output and DATA statements to refer to all its elements, but it is essential to know their order. The elements are referenced in the order of their positions in the computer's memory. For a one-dimensional array, the elements occur, as we might expect, in increasing order of their subscripts, but for multi-dimensional arrays, the ordering is less obvious. The rule is that the elements are ordered with the first subscript increasing most rapidly, then the next and so on, the last subscript increasing most slowly. Thus if MARK is declared as:INTEGER MARK(5,100)its elements are ordered in memory as shown above, and the statement:READ *,MARKis equivalent to:READ *, ((MARK(I,J),I=1,5),J=1,100)Of course, the order could be altered by swapping the control variables in the inner and outer implied DO loops thus:READ *, ((MARK(I,J),J=1,100),I=1,5)Note: if we consider a two dimensional array as a matrix, we should say that in memory its elements are stored column after colum. With the C language it should be row after row !We can use a two-dimensional array to solve the problem posed at the beginning of this section.
Example 2: write a program to read the marks of up to 100 students in five papers, and print, for each paper, the number of students sitting it, the average mark, and a list of the marks and their differences from the average. The marks are to be read as a list of five marks in the same order for each student, with a negative mark if the student did not sit a paper. The end of the data is indicated by a dummy mark of 999.
The outline is:
- Initialise the total mark for each of the five papers, a count of the number of students sitting it and a count of all the students.
- For up to 100 students, repeat:
- Read and store five marks
- If the first mark is 999, then continue from step 4.
- Otherwise increment the count of all students.
- For each of the five marks repeat:
- If the mark is not negative then:
- Increment the count of students sitting that paper.
- Add the mark to the total for that paper.
- If the mark is not negative then:
- Read a mark.
- If it is not 999 then:
- Print a message: 'Marks entered for more than 100 students.'
- STOP
- Repeat for each of five papers:
- If the count of students sitting the paper exceeds zero then:
- Compute the average mark for the paper.
- Print the number of the paper, the number of students sitting it, and the average mark.
- Print a list of all the marks in that paper and their differences from the average for the paper.
- For each student, repeat:
- If his/her mark in the paper is not negative, then:
- Print the mark.
- Compute and print the difference between the mark and the average for the paper.
- If his/her mark in the paper is not negative, then:
- Otherwise
- Print a message: 'No students sat paper number' paper_number
- If the count of students sitting the paper exceeds zero then:
Since the marks are read five subjects at a time for each student, it is convenient to store them in an array MARK(5,100). The program follows:
PROGRAM EXAM2 IMPLICIT NONE REAL AVMARK INTEGER MARK(5,100),TOTAL(5),COUNT(5),ALL,I,J,LAST DATA COUNT/5*0/,TOTAL/5*0/,ALL/0/ DO 20, J=1,100 READ *,(MARK(I,J),I=1,5) IF (MARK(1,J).EQ.999) GOTO 30 ALL = ALL+1 DO 10, I=1,5 IF (MARK(I,J).GE.0) THEN COUNT(I) = COUNT(I)+1 TOTAL(I) = TOTAL(I)+MARK(I,J) END IF 10 CONTINUE 20 CONTINUE READ *,LAST IF (LAST.NE.999) THEN PRINT *,'MARKS ENTERED FOR MORE THAN 100 STUDENTS.' STOP END IF 30 DO 50, I=1,5 IF (COUNT(I).GT.0) THEN AVMARK = 1.0*TOTAL(I)/COUNT(I) C MULTIPLY BY 1.0 TO CONVERT TO REAL AND AVOID TRUNCATION PRINT *,COUNT(I),' STUDENTS SAT PAPER NUMBER',I PRINT *,'THE AVERAGE MARK WAS', AVMARK PRINT *,'MARKS AND THEIR DIFFERENCES FROM THE AVERAGE:' DO 40, J=1,ALL IF (MARK(I,J).GE.0)PRINT *,MARK(I,J),MARK(I,J)-AVMARK 40 CONTINUE ELSE PRINT *,'NO STUDENTS SAT PAPER NUMBER',I END IF 50 CONTINUE ENDFigure 13: Exam marks program (version 2)
First page
-8- Input and output
- 8.1 The FORMAT statement
- 8.2 Formatted input
- 8.2.1 The I format descriptor
- 8.2.2 The F format descriptor
- 8.2.3 The E format descriptor
- 8.2.4 Repeat count
- 8.2.5 The X format descriptor
- 8.2.6 The T format descriptors
- 8.3 Formatted output
- 8.3.1 Vertical spacing
- 8.3.2 The I format descriptor
- 8.3.3 The F format descriptor
- 8.3.4 The E format descriptor
- 8.3.5 The literal format descriptors
- 8.4 More general input/output statements
- 8.5 The OPEN statement
- 8.6 Repetition of format specifications
- 8.7 Multi-record specifications
- 8.8 Sequential unformatted I/O
- 8.9 Direct access unformatted I/O
This chapter introduces input, output and format statements which give us greater flexibility than the simple READ and PRINT statements used so far.
A statement which reads information must:
- Scan a stream of information from an input device or file.
- Split the stream of information into separate items.
- Convert each item from its external form in the input to its internal (binary) representation.
- Store each item in a variable.
A statement which outputs information must:
- Retrieve each item from a variable or specify it directly as a constant.
- Convert each item from its internal form to an external form suitable for output to a given device or file.
- Combine the items with information required to control horizontal and vertical spacing.
- Send the information to the appropriate device or file.
The simple READ statement:
READ *, variable_list
reads a line (or record ) of information from the standard input (defined as the keyboard for programs run from a terminal) and stores it in the variables in variable_list. The asterisk refers to a list-directed format used to split the information into separate items using spaces and/or commas as separators and convert each item to the appropriate internal representation, which is determined by the type of the corresponding variable in variable_list.
Similarly, the simple PRINT statement:
PRINT *, output_list
uses a list-directed format to convert each constant, and the value of each variable, in output_list to a suitable form for output on standard output (defined for a program run from a terminal as the screen) and prints the list as a line of output, with spaces between the items.
8.1 The FORMAT statement
We can obtain greater control over the conversion and formatting of input/output items by replacing the asterisk in a READ or PRINT statement by the label of a FORMAT statement, for example:
READ 10,A,B,C 10 FORMAT(...)The FORMAT statement describes the layout of each item to be read or printed, and how it is to be converted from external to internal form or vice versa. It also describes the movements of an imaginary cursor which can be envisaged as scanning the input list. Its general form is:label FORMAT (specification_list)
label is a statement label. A FORMAT statement must always be labelled to provide a reference for use in input/output statements.
specification_list is a list of format descriptors (sometimes called edit descriptors), separated by commas. These describe the layout of each input or output item, and specify how it is to be converted (or edited) from external to internal form or vice versa.
FORMAT statements can be placed anywhere in a program. It is often convenient to place them all at the end (immediately before END), especially if some of them are used by more than one input/output statement.
8.2 Formatted input
The format descriptors used for input are summarised in Figure 14 and described in the following sections.
Descriptor Meaning
Iw Convert the next w characters to an INTEGER value. Fw.d Convert the next w characters to a REAL value. If no decimal point is included, the final d digits are the fractional part. Ew.d Convert the next w characters to a REAL value, interpreting them as a number in exponential notation. nX Skip the next n characters. Tc Skip to character absolute position c. TLn Skip to the character which is n characters to the left of the current character. TRn Skip to the character which is n characters to the right of the current character.
Figure 14: Some format descriptors for input8.2.1 The I format descriptor
This is used to read a value into an INTEGER variable. Its form is Iw, where w is an unsigned integer indicating the number of characters to be read (the width of the field). These characters must consist of decimal digits and/or spaces, which are interpreted as zeroes, with an optional + or - sign anywhere before the first digit. Any other characters will cause an input error.
Example:
READ 10,MEAN,INC 10 FORMAT(I4,I4)Input: b123b-5b(b represents a blank). This assigns a value of 123 to MEAN and -50 to INC.
8.2.2 The F format descriptor
This is used to read a value into a REAL variable. It has the form Fw.d, where w is an unsigned integer representing the width of the field and d is an unsigned integer representing the number of digits in the fractional part.
The corresponding input item must consist of decimal digits and/or spaces, with an optional sign anywhere before the first digit and an optional decimal point. As with the I format descriptor, spaces are interpreted as zeroes. If there is no decimal point in the item, the number of fractional digits is indicated by d. If the item includes a decimal point, d is ignored, and the number of fractional digits is as indicated.
Example:
READ 10,X,A,B,C,D 10 FORMAT(F4.5,F4.1,F2.2,F3.5,F3.0)Input: b1.5b123456789bbResults: X: 1.5 A: 12.3 B: 0.45 C: 0.00678 D: 900.0
8.2.3 The E format descriptor
This is used to read a value into a REAL variable. It has a similar form to the F format descriptor, but is more versatile, as it can be used to read input in exponential notation.
We saw in Chapter 2 that a REAL constant can be written in exponential notation as a REAL or INTEGER constant followed by an exponent in the form of the letter 'E' followed by the power of 10 by which the number is to be multiplied. For input, the exponent can also be a signed integer without the letter 'E'.
Example:
With a format descriptor of E9.2, all the following will be read as 1.26
0.126Eb01 1.26bEb00 1.26bbbbb 12.60E-01 bbb.126E1 bbbbbb126 126bbbbbb bbb12.6-1
8.2.4 Repeat count
The I, F and E format descriptors may be repeated by preceding them by a number indicating the number of repetitions. For example:
10 FORMAT(3I4)is equivalent to:10 FORMAT(I4,I4,I4)8.2.5 The X format descriptor
This is used with an unsigned integer prefix n to skip n characters.
Example:
READ 10,I,J 10 FORMAT(I4,3X,I3)Input: 123456789bResults: I: 1234 J: 890
8.2.6 The T format descriptors
The T (for tab), TL and TR format descriptors are used to move the cursor to a given position. This is defined absolutely by the T format descriptor or relative to the current position by the TL and TR descriptors.
Example:
READ 10,I,J,K 10 FORMAT(T4,I2,TR2,I2,TL5,I3)Input: 123456789bResults: I: 45 J: 89 K: 567
Notes:
- TRn is equivalent to nX.
- As illustrated by the example, tabs can be used not only to skip over parts of the input, but to go back and re-read parts of it.
- If TLn defines a position before the start of the record, the cursor is positioned at the first character. TL with a large value of n can therefore be used to return the cursor to the beginning of the record (as can T1).
8.3 Formatted output
Output statements use the same format descriptors as for input and another, the literal format descriptor, which is a string of characters for output. The descriptors are summarised in Figure 15 and described further in the following sections.
Descriptor Meaning Iw Output an INTEGER value in the next w character positions Fw.d Output a REAL value in the next w character positions, with d digits
in the fractional part.Ew.d Output a REAL value in exponential notation in the next w character positions,
with d digits in the fractional part.nX Skip the next n character positions. Tc Skip to character absolute position c. TLn Skip to the character which is n characters to the left of the current character. TRn Skip to the character which is n characters to the right of the current character. 'c1c2...cn' Output the string of n characters c1c2...cn starting at the next character position. nHc1c2...cn Output the string of n characters c1c2...cn starting at the next character position. Figure 15: Some format descriptors for output
8.3.1 Vertical spacing
As well as defining the layout of a line of output via an associated FORMAT statement, an output statement must define the vertical placement of the line on the screen or page of printed output. The method of doing this is described before the use of the format descriptors of Figure 15.
The computer uses the output list and the corresponding format specification list to build each line of output in a storage unit called an output buffer before displaying or printing it. When the contents of the buffer are displayed on the screen or printed on paper, the first character is not shown, but is interpreted as a control character, defining the vertical placement of the line. Four control characters are recognised, as shown in Figure 16.
Character Vertical spacing before output Space One line 0 (zero) Two lines 1 New page + No vertical spacing (i.e. current line is overprinted). Figure 16: Control characters for vertical spacing
The effect of any other character is not defined, but is usually the same as a space, i.e. output is on the next line.
Note: these vertical control characters are generally called "Fortran ASA carriage control characters". They are ineffective on modern displays and printers. To take them in account you must activate a filter which is contructor dependent. It converts them to equivalent ASCII control characters to simulate the vertical action.Fortran char. Equiv. ASCII control char. Hex. value Display Space Ignored . . 0 Line Feed '0A' ^J / Ctrl-J 1 Form Feed '0C' ^L / Ctrl-L + n Backspaces '08' ^H / Ctrl-H The overprint is simulated by as many backspaces as necessary to override the previous line: it's not always working very well!
On IBM, the available filter is called asa: for more details consult its man.
At IDRIS, a portable Fortran 95 program (prt_filter.f90) is available on our front-end SGI machine Rhodes to filter such an output. For more information, see /usr/local/public/src directory.
See also "The standard ASCII table" in chapter 14.
Incorrect output may be obtained if the control character is not taken into account. It is therefore best to use the format specification to insert a control character as the first character in a line, rather than to provide it via the output list. For example:
N = 15 PRINT 10,N 10 FORMAT(1X,I2)Buffer contents: b15Output: 15
The initial blank in the buffer is interpreted as a control character, and '15' is printed on the next line. However, if the FORMAT statement were:
10 FORMAT(I2)
the buffer contents would be '15'. On printing, the initial '1' would be interpreted as a control character, and '5' would be printed at the start of the next page.The following sections describe in more detail the effect of the format descriptors in output statements.
8.3.2 The I format descriptor
The format descriptor Iw is used to print an INTEGER value right-justified in a field of width w character positions, filling unused positions on the left with blanks and beginning with a '-' sign if the value is negative. If the value cannot be printed in a field of width w, the field is filled with asterisks and an output error is reported.
Example:
I = 15 J = 709 K = -12 PRINT 10,I,J,K, 10 FORMAT(1X,I4,I4,I4)Output: bb15b709b-12Notes:
- The first format descriptor 1X provides a space as
a control character to begin output on a new line. The next descriptor
I4 then prints the value 15 in a field of width 4. The
same effect could be obtained by using I5 as the first
descriptor, but it is clearer to use a separate descriptor for
the control character.
- The I, F and E format descriptors may be preceded by a repetition count r, where r is an unsigned integer. Thus rIw repeats the format descriptor Iw for r repetitions. For example, the above FORMAT statement could be replaced by:
10 FORMAT(1X,3I4)
8.3.3 The F format descriptor
The format descriptor Fw.d (F for floating point) is used to print a REAL value right-justified in a field of width w, with the fractional part rounded (not truncated) to d places of decimals. The field is filled on the left with blanks and the first non-blank character is '-' if the value is negative. If the value cannot be printed according to the descriptor, the field is filled with asterisks and an error is reported.
Example:
X = 3.14159 Y = -275.3024 Z = 12.9999 PRINT 10,X,Y,Z, 10 FORMAT(1X,3F10.3)Output: bbbbb3.142bb-275.302bbbb13.000The value of X is rounded up, that of Y is rounded down, and that of Z is rounded up, the 3 decimal places being filled with zeroes.
8.3.4 The E format descriptor
The format descriptor Ew.d is used to print a REAL value in exponential notation right-justified in a field of width w, with the fractional part rounded to d places of decimals. Thus the layout for a format descriptor of E10.3 is:
S0.XXXESXX <d> <---w---->
S indicates a position for a sign. The initial sign is printed only if negative, but the sign of the exponent is always printed. X indicates a digit.Example:
The value 0.0000231436 is printed as shown with the various format descriptors:
E10.4 0.2314E-04 E12.3 bbb0.231E-04 E12.5 b0.23144E-04
8.3.5 The literal format descriptors
The literal format descriptors 'c1c2...cn' and nHc1c2...cn place the string of n characters c1c2...cn directly into the buffer. Thus a PRINT statement using either of the following FORMAT statements will print the header: 'RESULTS' at the top of a new page:
10 FORMAT('1','RESULTS') 10 FORMAT(1H1,7HRESULTS)The quoted form is generally easier to use, but the 'H' form is convenient for providing control characters.
Note: the nHc1c2...cn Hollerith form is an obsolescent feature of Fortran 90 and deleted from Fortran 95.
A repetition count may be used with a literal format descriptor if the descriptor is enclosed in parentheses, e.g.
10 FORMAT(1X,3('RESULTS'))8.4 More general input/output statements
A record is a sequence of values or characters.
A file is a sequence of records.
An external file is one contained on an external medium (e.g. a magnetic disk).
Each Fortran input/output statement reads information from, or writes it to, a file. The file must be connected to an external unit, i.e. a physical device such as the keyboard or screen, or a magnetic disk. An external unit is referred to by a unit identifier, which may be:
- a non-negative integer expression, or:
- an asterisk ('*'), which normally refers to the keyboard for input, or the screen for output.
The READ and PRINT statements we have considered so far read from the file 'standard input', normally connected to the keyboard, and print on the file 'standard output', normally connected to the screen. To use different files and devices and to obtain various other options, we require a more general form of the READ statement for input, and a new statement, the WRITE statement for output. These statements have the form:
READ (cilist) input_list WRITE(cilist) output_listwhere cilist is a list of input-output specifiers, separated by commas. Each specifier takes the form:keyword = value
The specifiers may be in any order. In special cases noted below, only the value is required. Some of the keywords are:
UNIT FMT ERR END
The unit specifier must always be included. Its value must be a unit identifier, as defined above.
If the unit specifier is the first item in cilist, it may be denoted by its value only (without 'UNIT=').
Unit identifiers 5 and 6 are preconnected to the files 'standard input' and 'standard output' respectively.
The value of the format specifier FMT is the label of a FORMAT statement to be used for input/output conversion, or an asterisk to indicate list-directed formatting. A format specifier may be denoted by its value only (without 'FMT=') if it is the second item in cilist and follows a unit specifier also denoted by its value only.
Examples: if unit identifier 5 corresponds to standard input, the following are all equivalent:
READ(UNIT=5, FMT=100) X,Y,Z READ(FMT=100, UNIT=5) X,Y,Z READ(5, FMT=100) X,Y,Z READ(5, 100) X,Y,Z READ(*, 100) X,Y,ZAlso, the statements:READ(*,*) A,B,C READ(5,*) A,B,Care both equivalent to the list-directed input statement:READ *, A,B,CThe last two specifiers deal with special conditions. If an error occurs in input or output execution normally stops, but if an error specifier of the form:ERR = label
is included in cilist, execution continues from the statement labelled label. This makes it possible to include statements in the program to take special actions to deal with such errors.
If a READ statement tries to read more data than is available, an input error normally occurs. However, if a file ends with a special end-of-file record, a specifier of the form:
END = label
will cause execution to continue from the statement labelled label.
Note: the format specification list can be put inside the READ or WRITE statement. By example:
read(10, '(I4, F6.2)') K, X8.5 The OPEN statement
As noted above, the files 'standard input' and 'standard output' are preconnected to unit identifiers 5 and 6, and normally refer to the keyboard and screen, respectively. If other files, e.g. files on disk, are to be used, or if 'standard input' and 'standard output' are to be redefined, each file must be connected to an external unit by an OPEN statement, which has the form:
OPEN(openlist)where openlist is a list of specifiers of the form:
keyword = value
Specifiers may occur in any order. Two of the more important keywords are:
UNIT FILE
The unit specifier must be included. Its value must be a unit identifier.If the unit specifier is the first item in openlist, it may be denoted by its value only (without 'UNIT=').
The value of the file specifier is a character expression naming a file to be opened, i.e. connected to the external unit specified by the unit specifier. If the file does not exist, a new file is created.
Example:
OPEN(8, FILE='MYFILE.DAT')connects the file MYFILE.DAT to unit 8. If the file does not exist, it is created. READ and WRITE statements referring to this unit identifier will then read from or write to this file.8.6 Repetition of format specifications
If the number of items in an input or output list exceeds the number of format descriptors in the corresponding FORMAT statement, a new record is taken (a new line for terminal input/output) and the format specification list is re-used. This happens as often as required to deal with the complete list.
Example:
READ(5,10) A,B,C,P,Q,R,X,Y,Z 10 FORMAT(3F12.3)This reads three values from the first line of input into the variables A,B and C, from the second line into P,Q and R, and from the third line into X,Y and Z. Similarly:WRITE(6,10) A,B,C,P,Q,R,X,Y,Z 10 FORMAT(1X,3F12.3)prints the values of the variables three to a line on consecutive lines.The format specification list may also be re-used partially if it includes nested parentheses. The rules are:
- If there are no nested parentheses, the specification list is re-used from the beginning.
- If the list includes nested parentheses, the list is re-used from the left parenthesis corresponding to the last nested right parenthesis.
- If the left parenthesis so defined is preceded by a repeat count the list is re-used from immediately before the repeat count.
This is illustrated by the following examples, in which a vertical bar indicates the point from which repetition, if required, begins:
10 FORMAT(I6, 10X,I5, 3F10.2) |--------------------- 20 FORMAT(I6, 10X,I5, (3F10.2)) |------- 30 FORMAT(I6, (10X,I5), 3F10.2) |---------------- 40 FORMAT(F6.2, (2F4.1,2X,I4, 4(I7,F7.2))) |------------------------- 50 FORMAT(F6.2, 2(2F4.1,2X,I4), 4(I7,F7.2)) |--------- 60 FORMAT(F6.2,(2(2F4.1,2X,I4), 4(I7,F7.2))) |---------------------------8.7 Multi-record specifications
Repetitions can be used as in the last section to read in or print out a sequence of values on consecutive lines using the same format specification list. It is also useful to be able to specify the format of several consecutive lines (or records) in a single format specification. This can be done using the / format descriptor, which marks the end of a record. Unlike other format descriptors, / need not be preceded or followed by a comma.
On input, / causes the rest of the current record to be skipped, and the next value to be read from the first item on the next record. For example:
READ(5,100) A,B,C,I,J,K 100 FORMAT(2F10.2,F12.3/I6,2I10)reads three REAL values into A,B and C from a line, ignores anything more on that line, and reads three INTEGER values into I,J and K from the next line.Consecutive slashes cause records to be skipped. Thus if the FORMAT statement in the above example were changed to:
100 FORMAT(2F10.2,F12.3//I6,2I10)
a complete line would be skipped before the values were read into I,J and K.On output, a / marks the end of a record, and starts a new one. Consecutive slashes cause blank records to be output. For example:
WRITE(6,200) A,B,A+B,A*B 200 FORMAT(1H1////T10,'MULTIPLE LINES EXAMPLE'/// * 1X,'THE SUM OF',F5.2,' AND',F5.2,' IS',F5.2/ * 1X,'AND THEIR PRODUCT IS',F8.2////)prints four blank lines and a header at the top of a new page, followed by two blank lines, then the sum and product on consecutive lines followed by four blank lines.The following example illustrates the use of formatting to produce output in tabular form with headers and regular spacing.
Example 1: rewrite the degrees to radians conversion program (Chapter 6, Example 2) to print angles from 1 to 360 degrees in 1 degree intervals and their equivalents in radians. The results should be printed 40 lines to a page, with the values suitably formatted, blank lines separating groups of 10 consecutive lines, headers for the 'Degrees' and 'Radians' columns, and a header and page number at the start of each page.
The outline is:
- Initialise the page number to zero.
- Compute the conversion factor.
- Repeat for angles in degrees from 1 to 360 in steps of 1:
- If the angle in degrees is one more than a multiple of 40 then:
- Increment the page number.
- Print a page header, page number and column headers, followed by a blank line.
- Otherwise, if the angle in degrees is one more than a multiple
of 10, then:
- Print a blank line.
- Compute the angle in radians.
- Print the angle in degrees and radians.
- If the angle in degrees is one more than a multiple of 40 then:
and the program follows:
PROGRAM ANGLES IMPLICIT NONE REAL RADIAN,CONFAC INTEGER DEGREE,PAGENO,OUT,N DATA OUT/6/ C UNIT NUMBER FOR OUTPUT. A DIFFERENT DEVICE COULD BE USED BY C CHANGING THIS VALUE DATA PAGENO/0/ CONFAC = 3.141593/180.0 C CONVERSION FACTOR FROM DEGREES TO RADIANS DO 10, DEGREE = 1,360 N = DEGREE-1 IF (N/40*40 .EQ. N) THEN C PRINT PAGE HEADER, NUMBER AND COLUMN HEADERS PAGENO = PAGENO+1 WRITE(OUT,100)PAGENO ELSE IF (N/10*10 .EQ.N) THEN WRITE(OUT,110) END IF RADIAN = DEGREE*CONFAC WRITE(OUT,120)DEGREE,RADIAN 10 CONTINUE 100 FORMAT(1H1//1X,'DEGREES TO RADIANS CONVERSION TABLE', * T74,'PAGE',I2//1X,'DEGREES RADIANS'/) 110 FORMAT(1X) 120 FORMAT(1X,I5,T10,F7.5) ENDFigure 17: Degrees to radians conversion program (version 3)8.8 Sequential unformatted I/O
This program writes one record containing the array T. The data are written without any format specification (often said "binary" mode). After having rewound it, this record is read.Example:
REAL T(100) DO 10 I=1, 100 T(I) = SIN(REAL(100)/123.) 10 CONTINUE OPEN(UNIT=10, FILE='MYFIC', ACCESS='SEQUENTIAL', 1 FORM='UNFORMATTED', STATUS='NEW') WRITE(UNIT=10) T REWIND 10 READ(UNIT=10) T CLOSE(UNIT=10) ENDNote: some parameters of the OPEN statement are set by defaut :
ACCESS='SEQUENTIAL' FORM='UNFORMATTED' STATUS='UNKNOWN'
It could have been coded more shortly as:OPEN(10, FILE='MYFIC', STATUS='NEW')8.9 Direct access unformatted I/O
This program writes five records, each of then containing a part of the array T. The data are written without any format specification (often said "binary" mode).Then it reads the third record and stocks it in array TR.Example:
REAL T(100), TR(20) DO 10 I=1, 100 T(I) = REAL(I) 10 CONTINUE OPEN(UNIT=10, FILE='MYFIC', ACCESS='DIRECT', FORM='UNFORMATTED', 1 STATUS='UNKNOWN', RECL=80) DO 20 K=1, 5 WRITE(UNIT=10, REC=K) (T(I), I=(K-1)*20+1, K*20) 20 CONTINUE READ(UNIT=10, REC=3) TR PRINT *, TR ENDNotes:
- REC is the record number: that's the number of
the record to be read or written.
- RECL is the record length: it's value specifies the length (in bytes) of each record. Here REC=20*4 bytes.
First page
-9- Functions and subroutines
- 9.1 Intrinsic functions
- 9.2 External functions
- 9.2.1 The FUNCTION statement
- 9.2.1.1 Type
- 9.2.1.2 The argument list
- 9.2.2 The function reference
- 9.2.2.1 Actual and dummy arguments
- 9.2.3 Evaluation of a function
- 9.2.4 Examples
- 9.2.1 The FUNCTION statement
- 9.3 Statement functions
- 9.3.1 Rules
- 9.4 Subroutines
- 9.5 Procedures as arguments
- 9.6 Local variables
Very often, a program has to perform a computation several times using different values, producing a single value each time. An example is the conversion of an angle in degrees to an equivalent in radians in Example 1 of the previous chapter.
In Fortran, such a computation can be defined as a function and referred to by a name followed by a list of the values (called arguments) which it uses, in parentheses, i.e.
name([argument_list])
where argument_list is an optional list of arguments separated by commas. Note that the parentheses must be included even if argument_list is omitted, i.e.
name()
Such a function reference can be used in the same way as a variable or array element, except that it cannot be the object of an assignment. Like a variable or array element, a function reference is evaluated and the value obtained is substituted for it in the expression in which it appears. The type of a function is the type of the value so obtained.
Thus, in the above example, a REAL function DGTORD might be defined to convert an angle in degrees to an equivalent in radians. The function would have a single argument, of type INTEGER, representing the value of the angle in degrees, and would be evaluated to obtain the equivalent in radians. The function might be used in an assignment statement like:
RADIAN = DGTORD(DEGREE)The definition of a function must include a definition of its type and the number and types of its arguments. In a function reference the number and type of the arguments must be as defined. Thus, for example:RADIAN = DGTORD(DEGREE,X)would be an error.As the above example illustrates, a function reference has an identical form to an array element, and may be used in a similar context. Fortran distinguishes between the two by checking whether the name has been declared as an array, and assuming that it is a function if it has not. Thus, for example, if DGTORD were declared as:
REAL DGTORD(100)then DGTORD(DEGREE) would be interpreted as an array element and not a function reference.
9.1 Intrinsic functions
Fortran provides a wide range of intrinsic functions, which are defined as part of the language. Many of them have an argument, or list of arguments, which may be of different types in different references. Most, though not all, of these return a value of the same type as that of their arguments in any reference. For example, the function ABS returns the absolute value of its argument, which may be REAL or INTEGER. Thus
ABS(X)returns the absolute value of the REAL variable X as a REAL value, while
ABS(N)returns the absolute value of the INTEGER variable N as an INTEGER value.
A function of this kind is called a generic function. Its name really refers to a group of functions, the appropriate one being selected in each reference according to the type of the arguments.
Figure 18 is a list of some of the more frequently used intrinsic functions. I and R indicate INTEGER and REAL arguments respectively. Where an argument represents an angle, it must be in radians.
Name Type Definition
ABS(IR) Generic Absolute value: |IR| ACOS(R) REAL arccos(R) AINT(R) REAL Truncation: REAL(INT(R)) ANINT(R) REAL Nearest whole number: REAL(INT(R+0.5)) if R>=0 REAL(INT(R-0.5)) if R<=0 ASIN(R) REAL arcsin(R) ATAN(R) REAL arctan(R) COS(R) REAL cos(R) COSH(R) REAL cosh(R) DIM(IR1,IR2) Generic Positive difference: MAX(IR1-IR2,0) EXP(R) REAL eR INT(R) INTEGER INTEGER portion of R LOG(R) REAL Natural logarithm: logeR LOG10(R) REAL Common logarithm: log10R MAX(IR1,IR2,...) Generic Largest of IR1,IR2,... MIN(IR1,IR2,...) Generic Smallest of IR1,IR2,... MOD(IR1,IR2) Generic Remainder: IR1-INT(IR1/IR2)*IR2 NINT(R) INTEGER Nearest integer: INT(ANINT(R)) REAL(I) REAL Real equivalent of I SIGN(IR1,IR2) Generic Transfer of sign: |IR1| if IR2>=0 -|IR1| if IR2<0 SIN(R) REAL sin(R) SINH(R) REAL sinh(R) SQRT(R) REAL R TAN(R) REAL tan(R) TANH(R) REAL tanh(R)
Figure 18: Some common intrinsic functions9.2 External functions
As well as using the intrinsic functions provided by the language, a programmer may create and use his/her own external functions. These functions may be included in the same source file as a program which uses them and compiled along with it, or may be written and compiled separately to obtain separate object files which are then linked to the object version of the program to obtain an executable program, in the same way as the library subprograms shown in Figure 3 on page 2. In either case, the program and functions are entirely independent program units.
A Fortran source file consists of one or more program units in any order. One of these may be a main program unit, which begins with an optional PROGRAM statement and ends with an END statement. The others are subprograms, which may be external functions or subroutines. (Subroutines are explained later in the chapter.)
An external function program unit begins with a FUNCTION statement and ends with an
Figure 19 illustrates a Fortran source file containing three program units, a main program MAIN and two functions FUN1 and FUN2. The order of the program units is immaterial.
PROGRAM MAIN . . . . . . . . END FUNCTION FUN1(arg1,...) . . . . . . . . END FUNCTION FUN2(arg1,...) . . . . . . . . ENDFigure 19: A Fortran source file containing two functionsProvided that the program MAIN includes no references to any other external functions, the file could be compiled, and the resulting object file linked with the library subprograms to obtain an executable program.
The functions might also be placed in one or two separate files and compiled separately from the main program. The object file or files thus obtained could then be linked with the library subprograms and the object version of the program MAIN or any other program containing references to them. In this way a programmer can create his/her own subprogram libraries for use by any program.
9.2.1 The FUNCTION statement
As shown above, an external function must begin with a FUNCTION statement. This has the form:
[type] FUNCTION name([argument_list])As before, square brackets indicate that an item is optional.
9.2.1.1 Type
Each function has a type corresponding to the type of value returned by a reference to it. As for variables, the type of a function may be specified explicitly or assigned implicitly according to the first letter of the function name. For example, the function:
FUNCTION FUN1(arg1,...)returns a value of type REAL, butINTEGER FUNCTION FUN1(arg1,...)returns a value of type INTEGER.If the type of a function differs from that implied by the first letter of its name, it must be declared in a type specification in any program which refers to it. Thus any program using the second version of FUN1 above would include the name FUN1 in an INTEGER type specification statement, e.g.
INTEGER FUN19.2.1.2 The argument list
argument_list is an optional list of dummy arguments, separated by commas. Each dummy argument is a name similar to a variable or array name, which represents a corresponding actual argument used in a function reference. Dummy arguments, and variables used in a function, are defined only within it. They may therefore be identical to variable or array names used in any other program unit.
If a dummy argument represents an array, it must appear in a type specification or DIMENSION statement in the function. If it represents a variable, it may appear in a type specification, or may be typed by default.
Example:
FUNCTION FUN1(A,B,N) REAL A(100) INTEGER BHere, A represents a REAL array of dimension 100, and B and N represent INTEGER variables.A function may have no arguments, e.g.
FUNCTION NOARGS()9.2.2 The function reference
As we have seen, a function reference has the form:
name(argument_list)
argument_list is a list of actual arguments, which must match the list of dummy arguments in the FUNCTION statement with respect to the number of arguments and the type of each argument. For example:
REAL X(100) . . . RESULT = FUN1(X,J,10)would be a valid reference to the function FUN1(A,B,N) shown above.If a dummy argument is a variable name, the corresponding actual argument may be any expression of the same type, i.e. a constant, variable, array element or more complex arithmetic expression.
If a dummy argument is an array name, the actual argument may be an array or array element. The dimensions of the dummy array may be variable if they are also dummy arguments.
Example:
REAL X(5,10) . . . Y = FUN(X,5,10) . . . END FUNCTION FUN(A,M,N) REAL A(M,N) . . .9.2.2.1 Actual and dummy arguments
The dummy arguments and corresponding actual arguments provide a means of exchanging information between a program unit and a function.
Each actual argument refers to a word or other unit of storage. However, no storage is reserved for a dummy argument; it is simply a name. When a function reference is evaluated, the address of each actual argument is passed to the function, and the corresponding dummy argument is set to refer to it. The dummy argument may therefore be used in the function as a variable or array referring to the same unit of storage as the actual argument.
Thus if a dummy argument represents a variable, its value on entry to the function is that of the corresponding actual argument when the function is referenced. If its value is changed in the function by an assignment or READ statement, the actual argument will be correspondingly changed after the function reference has been evaluated.
Arrays as arguments
If a dummy argument is an array, the corresponding actual argument may be an array or array element. In the former case, the elements of the dummy array correspond to the elements of the actual array in the order of their storage in memory. This, however, does not imply that the subscripts are identical, or even that the two arrays have the same number of subscripts. For example, suppose that the function:
FUNCTION FUN(A) REAL A(9,6) . . . ENDis referenced by program MAIN as follows:PROGRAM MAIN REAL X(100),Y(0:5,-10,10) . . . F1 = FUN(X) F2 = FUN(Y) . . . ENDThen the correspondence between some elements of the dummy array A and the actual arrays X and Y in the two function references is as shown below:A(1,1) X(1) Y(0,-10) A(6,1) X(6) Y(5,-10) A(7,1) X(7) Y(0,-9) A(1,2) X(10) Y(3,-9) A(5,4) X(32) Y(1,-5) A(9,6) X(54) Y(5,-2)
If the actual argument is an array element, the first element of the dummy array corresponds to that element. Thus, if the function references:F3 = FUN(X(15)) F4 = FUN(Y(3,0))were included in the program above, the following items would correspond in the two references:A(1,1) X(15) Y(3,0) A(4,1) X(18) Y(0,1) A(9,1) X(23) Y(5,1) A(1,2) X(24) Y(0,2) A(5,4) X(46) Y(4,5) A(9,6) X(68) Y(2,9)
Such complicated relationships between actual and dummy arguments can sometimes be useful, but are in general best avoided for reasons of clarity.9.2.3 Evaluation of a function
Once the dummy arguments have been initialised as described above, the statements comprising the body of the function are executed. Any statement other than a reference to the function itself may be used. At least one statement must assign a value to the function name, either by assignment, or less commonly, by a READ statement. Execution of the function is stopped, and control returned to the program unit containing the function reference, by a RETURN statement, written simply as:
RETURNThe value of the function name when RETURN is executed is returned as the function value to the program unit containing the function reference.
We can now write the function DGTORD suggested at the beginning of the chapter, to convert an INTEGER value representing an angle in degrees, to a REAL value representing the equivalent in radians. Our function uses the intrinsic function ATAN to compute the conversion factor.
FUNCTION DGTORD(DEG) INTEGER DEG CONFAC = ATAN(1.0)/45.0 DGTORD = DEG*CONFAC RETURN ENDAs a second example, the following function returns the mean of an array of N real numbers.REAL FUNCTION MEAN(A,N) REAL A(N) SUM = 0.0 DO 10, I=1,N SUM = SUM+A(I) 10 CONTINUE MEAN = SUM/N RETURN ENDNote that, since the type of this function differs from that implied by the first letter of its name, any program referring to it must declare the name in a type specification, e.g.REAL MEAN9.3 Statement functions
If a function involves only a computation which can be written as a single statement, it may be declared as a statement function in any program unit which refers to it. The declaration has the form:
name(argument_list) = expression
where:
name is the name of the statement function. argument_list is a list of dummy arguments. expression is an expression which may include constants, variables and array elements defined in the same program unit, and function references.The declaration must be placed after all type specifications, but before the first executable statement.Thus the function DGTORD might be declared as a statement function in the program ANGLES:
DGTORD(DEGREE) = DEGREE*ATAN(1.0)/45.0Note: statement functions are obsolete with the Fortran 95 standard. They are replaced by internal functions (see "obsolescent features" in Fortran 95 manuals).
The name of a statement function must be different from that of any variable or array in the same program unit.
The type of a statement function may be specified explicitly in a separate type specification or determined implicitly by the first letter of its name.
A dummy argument may have the same name as a variable or array in the same program unit. If so, it has the same type as the variable or array but is otherwise distinct from it and shares none of its attributes. For example, in the program ANGLES, the dummy argument DEGREE of the statement function DGTORD has the same name as the variable DEGREE declared in the program, and therefore has the correct (INTEGER) type, but is a different entity. If the program included the declaration:
INTEGER DEGREE(100)the dummy argument DEGREE would be an INTEGER variable, not an array.If a dummy argument does not have the same name as a variable or array in the same program unit, it is typed implicitly according to its first letter, e.g.
DGTORD(IDEG) = IDEG*ATAN(1.0)/45.0expression may include references to functions, including statement functions. Any statement function must have been previously defined in the same program unit.A SUBROUTINE is a subprogram similar in most respects to a function. Like a function, a subroutine has a list of dummy arguments used to exchange information between the subroutine and a program unit referring to it. Unlike a function, a subroutine does not return a value via its name (and therefore has no type), but it may return one or more values via its arguments.
A subroutine subprogram begins with a SUBROUTINE statement and ends with END. The SUBROUTINE statement has the form:
SUBROUTINE name[(argument_list)]where name and argument_list have the same meanings as in the FUNCTION statement. The square brackets indicate that the item (argument_list) is optional, i.e. a subroutine may have no arguments, in which case the SUBROUTINE statement is simply:SUBROUTINE nameAs for a function, a subroutine must include at least one RETURN statement to return control to the program unit referring to it.A subroutine is referenced by a CALL statement, which has the form:
CALL name[(argument_list)]where argument_list is a list of actual arguments corresponding to the dummy arguments in the SUBROUTINE statement. The rules governing the relationship between actual and dummy arguments are the same as for functions.Functions (intrinsic and external) and subroutines are often called procedures.
In Example 1 of Chapter 8, the steps required to print a page header and column headers at the top of each page might be written as a subroutine. The steps are:
- Increment the page number.
- Print a page header, page number and column headers, followed by a blank line.
The subroutine therefore has two dummy arguments, one representing the page number and the other representing the output device, and includes the WRITE statement and FORMAT statements required to print the page and column headers. The subroutine follows:
SUBROUTINE HEADER(PAGENO,OUTPUT) C PRINT PAGE HEADER, NUMBER AND COLUMN HEADERS INTEGER PAGENO,OUTPUT PAGENO = PAGENO+1 WRITE(OUTPUT,100)PAGENO 100 FORMAT(1H1//1X,'DEGREES TO RADIANS CONVERSION TABLE', * T74,'PAGE',I2//1X,'DEGREES RADIANS'/) RETURN ENDNote that the argument OUTPUT is used to receive a value from the calling program, while PAGENO both receives and returns a value.The degrees to radians conversion program can now be rewritten using the subroutine HEADER and function DGTORD as follows:
PROGRAM ANGLES INTEGER DEGREE,PAGENO,OUT DATA OUT/6/ C UNIT NUMBER FOR OUTPUT. A DIFFERENT DEVICE COULD BE USED BY C CHANGING THIS VALUE DATA PAGENO/0/ DO 10, DEGREE = 1,360 N = DEGREE-1 IF (N/40*40 .EQ. N) THEN CALL HEADER(PAGENO,OUT) ELSE IF (N/10*10 .EQ.N) THEN WRITE(OUT,110) END IF WRITE(OUT,120)DEGREE,DGTORD(DEGREE) 10 CONTINUE 110 FORMAT(1X) 120 FORMAT(1X,I5,T10,F7.5) ENDFigure 20: Degrees to radians conversion program (version 4)A program unit can pass the names of procedures as arguments to a function or subroutine. The calling program unit must declare these names in an EXTERNAL statement for external procedures (functions or subroutines), or INTRINSIC statement for intrinsic functions. The statements have the form:
EXTERNAL list INTRINSIC listrespectively, where list is a list of external procedures, or intrinsic functions respectively.
If an actual argument is a procedure name, the corresponding dummy argument may be:
- used as a procedure in a CALL statement or function reference, or:
- passed as an actual argument to another procedure. In this case, it must be listed in an EXTERNAL statement.
In this way, a procedure name can be passed from one procedure to another for as many levels as required.
Example 1
In Figure 21, the program MAIN passes the names of the subroutine ANALYS and the intrinsic function SQRT as actual arguments to the subroutine SUB1, corresponding to its dummy arguments SUB and FUN respectively. In SUB1, SUB appears in a CALL statement in which it is replaced in this instance by a call of ANALYS, while FUN appears in an EXTERNAL statement and is passed as an actual argument to SUB2, corresponding to its dummy argument F. In SUB2, F appears followed by a left parenthesis. Because F is not declared as an array, this is interpreted as a function reference, and is replaced by a reference to SQRT.
Note that although SQRT is an intrinsic function and is declared as such in program MAIN, FUN, the corresponding dummy argument of subroutine SUB1, is declared in SUB1 as EXTERNAL because FUN is a dummy procedure name corresponding to a function defined externally to SUB1.
PROGRAM MAIN EXTERNAL ANALYS INTRINSIC SQRT . . . CALL SUB1(ANALYS,SQRT,A,B) . . . . END SUBROUTINE SUB1(SUB,FUN,X,Y) EXTERNAL FUN . . . CALL SUB(...) . . . CALL SUB2(FUN,X,Y) . . . END SUBROUTINE SUB2(F,P,Q) . . . Q = F(P) . . . ENDFigure 21: Procedures as argumentsExample 2
In Figure 22, the subroutine TRIG has three dummy arguments, X representing an angle in radians, F representing a trigonometric function, and Y representing that function of X. The main program includes four calls to TRIG, using the intrinsic functions SIN, COS and TAN and the external function COT, which computes the cotangent.
PROGRAM MAIN EXTERNAL COT INTRINSIC SIN,COS,TAN . . . CALL TRIG(ANGLE,SIN,SINE) . . . CALL TRIG(ANGLE,COS,COSINE) . . . CALL TRIG(ANGLE,TAN,TANGT) . . . CALL TRIG(ANGLE,COT,COTAN) . . . END SUBROUTINE TRIG(X,F,Y) Y = F(X) RETURN END FUNCTION COT(X) COT = 1.0/TAN(X) RETURN ENDFigure 22: Subroutine to compute any trigonometric function.9.6 Local variables
The variables used in a subprogram, other than its arguments, are local variables, defined only within it, and therefore distinct from any identically named variables used elsewhere. When a RETURN statement is executed, they become undefined, and their addresses may be used by other program units. Therefore, if a subprogram is executed several times, the values of its local variables are not preserved from one execution to the next.
The values of local variables can be preserved by a SAVE statement, which has the form:
SAVE [variable_list]where variable_list is a list of local variables, separated by commas. The statement causes the values of all variables in variable_list to be saved. If variable_list is omitted, the values of all local variables are saved.SAVE is a non-executable statement and must be placed before the first executable statement or DATA statement.
Example: each time the following function is executed, it prints a message indicating how many times it has been referenced.
FUNCTION AVE(X,Y) INTEGER COUNT SAVE COUNT DATA COUNT/0/ COUNT = COUNT+1 WRITE(6,10)COUNT . . . 10 FORMAT(1X,'FUNCTION AVE REFERENCED',I3,' TIMES.') . . . ENDFirst page
-10- The type CHARACTER
- 10.1 CHARACTER constants
- 10.2 CHARACTER variables
- 10.2.1 Arrays
- 10.2.2 Assignment
- 10.3 CHARACTER expressions
- 10.3.1 Concatenation
- 10.3.2 Extraction of a substring
- 10.4 Input and output
- 10.5 Logical expressions
Constants of type CHARACTER were briefly introduced in Chapter Two. You will recall that a CHARACTER constant (or string) is a sequence of characters delimited by single quotes, and that single quotes may be included by writing two consecutively.
10.2 CHARACTER variables
Variables of type CHARACTER must be declared in a CHARACTER type specification, which specifies the length of each variable (i.e. the number of characters it contains). This takes the form:
CHARACTER[*len] var[*vlen] [,var[*vlen]] ...len and vlen are unsigned INTEGER constants or constant expressions in parentheses.var is a variable name.
Thus the simplest form of declaration is:
CHARACTER var [,var]...which specifies that each CHARACTER variable var contains one characterThe form:
CHARACTER*len var [,var]...specifies that each CHARACTER variable var contains len characters.The form:
CHARACTER var[*vlen] [,var[*vlen]]...may be used to specify a different length vlen for each variable var. If *vlen is omitted, one character is assigned.Finally, the form:
CHARACTER*len var[*vlen] [,var[*vlen]]...specifies that if a variable var is followed by *vlen it contains vlen characters, and otherwise it contains len characters.Example: the following specification assigns a length of 4 characters to the CHARACTER variables A and C, and 6 characters to B.
CHARACTER*4 A,B*6,C10.2.1 CHARACTER arrays
Example:
CHARACTER*4 A(3,4),B(10,20)*6This declares two CHARACTER arrays: A with 12 elements each 4 characters long, and B with 200 elements each 6 characters long.A CHARACTER variable may only be assigned a value of type CHARACTER.
If the length of a variable differs from that of the value assigned to it, the following rules are applied:
- If the length of the value is less than that of the variable, it is extended on the right with blanks.
- If the length of the value is greater than that of the variable, it is truncated on the right.
Example:
PROGRAM CHAREX CHARACTER*4 A*3,B,C A = 'END' B = A C = 'FINAL' STOP ENDResults:ValueFigure 23: Character assignment
A 'END' B 'END ' C 'FINA'10.3 CHARACTER expressions
Two operations are defined for character strings: concatenation and extraction of a substring.
The concatenation operator // joins two character string operands together in sequence.
Example:
If A is a CHARACTER variable of length 5, the assignment:
A = 'JIM'//'MY'assigns a value of 'JIMMY' to A.10.3.2 Extraction of a substring
A substring is a string of contiguous characters forming part of another string. A substring is extracted by writing a CHARACTER variable followed by one or two INTEGER expressions in parentheses and separated by a colon, indicating the leftmost and rightmost character positions of the substring in the larger string. If the first expression is omitted, the substring begins at the beginning of the string. If the second is omitted, it ends at the end of the string.
Example:
If the CHARACTER variable LANG has the value 'FORTRAN', some substrings are:
Substring Value
A substring reference can be used in the same way as a CHARACTER variable. Thus part of a string can be changed by an assignment to a substring.
LANG(1:1) 'F' LANG(1:7) 'FORTRAN' LANG(2:3) 'OR' LANG(7:7) 'N' LANG(:4) 'FORT' LANG(5:) 'RAN'Example:
The following assignment will change the value of the CHARACTER variable LANG from 'FORTRAN' to 'FORMATS':
LANG(4:7) = 'MATS'10.4 Input and output
When a CHARACTER variable is used in a list directed input statement, the value read must be delimited by single quotes. These are required because the value may include characters such as blanks, commas or slashes (/), which are normally recognised as separators between input items.
When a CHARACTER expression is used in a list directed output statement, it is printed in full using as many character positions as required. This form of output has been used in earlier program examples, e.g.
PRINT *,'THIS IS A STRING.'Character strings can be used in formatted input and output with the A format descriptor, which has the form A or Aw, where w is the field width in characters. The effect for input and output is shown in Figure 24.Descriptor Input Output Aw Input w characters. Output characters in the next
w character positions.A Input sufficient characters Output the output list item
to fill the input list item
with no leading or trailing
blanks.Figure 24: The A format descriptor
If w differs from the length len of the input or output item, the rules are:
For input:
- If w is less than len then blanks are added on the right to fill the input list item. This is similar to assignment.
- If w is greater than len then the right-most len characters of the data item are stored in the input list item. This is the opposite of what happens in assignment.
For output:
- If w is less than len then the left-most w characters will be output.
- If w is greater than len then the string is right-justified in the output field and extended on the left with blanks.
These rules ensure consistency of input and output. If a string is written out, and the result read using the same format, the value read in will be the same as that originally written out. This would not be so, for example, if rule (ii) for input were changed to store the left-most len characters as for assignment. This is illustrated in Figure 25, in which the output of the program CHROUT is read by the program CHRIN.
PROGRAM CHROUT CHARACTER*4 A,B A = 'WHAT' B = 'FOR ' WRITE(2,100)A,B 100 FORMAT(1H ,A6,3X,A3) STOP ENDOutput:bbWHATbbbFOR PROGRAM CHRIN CHARACTER*4 A,B READ(1,200)A,B 200 FORMAT(A6,3X,A3) STOP ENDResult:A contains 'WHAT'. B contains 'FORb'.
(b represents a blank.)
Figure 25: Character input and output
10.5 Logical expressions
Character strings can be used in logical expressions with the six relational operators .GT., .GE., .EQ., .NE., .LE. and .LT.. The definition of the operators depends on the coding scheme used to represent characters in binary form, which can be used to define a collating sequence of all valid characters in order of their binary codes. Two coding schemes, ASCII and EBCDIC, are in common use. The two collating sequences are different but have the following rules in common:
- Letters are in alphabetic sequence from A to Z.
- Digits are in sequence from 0 to 9.
- The sequence of digits either precedes or follows the sequence of letters; there is no overlapping.
- The blank character is the first in the sequence.
Relational expressions with single character operands are defined with reference to the collating sequence. For example, if CHAR1 and CHAR2 are two CHARACTER variables of length 1, then CHAR1.GT.CHAR2 evaluates to .TRUE. if CHAR1 comes after CHAR2 in the collating sequence. The other operators are similarly defined.
A relational expression with two character string operands of any length is evaluated in the following stages:
- If the operands are of unequal length, the shorter one is
extended on the right with blanks to correspond in length with
the longer.
- Corresponding characters in the two operands are compared
using the relational operator, starting from the left, until:
- A difference is found. The value of the relationship between
the operands is that between the two differing characters.
or: - The end of the operands is reached. Any expression involving equality evaluates as .TRUE. and any other as .FALSE.
- A difference is found. The value of the relationship between
the operands is that between the two differing characters.
Examples:
'ADAM'.GT.'EVE' evaluates to .FALSE. because 'A' precedes 'E' in the collating sequence.
'ADAM'.LE.'ADAMANT' evaluates to .TRUE. 'ADAM' is extended on the right with blanks, the first four characters are found to be identical, and the expression:
' '.LE.'A' then evaluates to .TRUE..The value of such expressions as:
'XA'.LT.'X4' 'VAR-1'.LT.'VAR.1'is undefined by the collating rules of Figure 26. In the first example, the rules do not stipulate whether letters come before or after digits, while in the second example, the characters '-' and '.' are not included in the rules. The value of such expressions depends on the coding scheme used by the computer system.First page
-11- Additional information types
- DOUBLE PRECISION
- 11.1 DOUBLE PRECISION constants
- 11.1.2 DOUBLE PRECISION variables
- 11.1.3 Input and output
- 11.1.4 Expressions
- 11.1.5 Functions
- 11.2 COMPLEX
- 11.2.1 COMPLEX constants
- 11.2.2 COMPLEX variables
- 11.2.3 Input and output
- 11.2.4 COMPLEX expressions
- 11.2.5 COMPLEX functions
A REAL variable or constant occupies one word of storage and this limits its accuracy. When greater accuracy is required, DOUBLE PRECISION variables and constants may be used. These occupy two words of storage and can store a greater number of significant digits.
11.1.1 DOUBLE PRECISION constants
DOUBLE PRECISION constants are written in exponential form, but with the letter 'D' in place of 'E', e.g.
1D-7 14713D-3 12.7192D0 9.413D5
11.1.2 DOUBLE PRECISION variables
DOUBLE PRECISION variables must be declared in a type specification of the form:
DOUBLE PRECISION variable_listwhere variable_list is a list of variables, separated by commas.DOUBLE PRECISION values can be used in list-directed input and output in the same way as REAL values. In formatted input and output, they may be used with the F and E format specifications and with a new format specification D, which has a similar form to the E specification, i.e.
Dw.d
In output, this specification prints a value in exponential form with a 'D' instead of an 'E'.
If both operands of an arithmetic operation are of type DOUBLE PRECISION, the result is also of type DOUBLE PRECISION. If one operand is of type REAL or INTEGER, the result is of type DOUBLE PRECISION, but this does not imply that the other operand is converted to this type.
All the intrinsic functions in Figure 18 on page 45 which take REAL arguments also take DOUBLE PRECISION arguments and return DOUBLE PRECISION values.
Fortran provides for the representation of complex numbers using the type COMPLEX.
A COMPLEX constant is written as two REAL constants, separated by a comma and enclosed in parentheses. The first constant represents the real, and the second the imaginary part.
Example:
The complex number 3.0-i1.5, where i²= -1, is represented in Fortran as:
(3.0,-1.5).
COMPLEX variables must be declared in a COMPLEX type specification:
COMPLEX variable_listIn list-directed output, a COMPLEX value is printed as described under 'COMPLEX constants'. In list-directed input, two REAL values are read for each COMPLEX variable in the input list, corresponding to the real and imaginary parts in that order.
In formatted input and output, COMPLEX values are read or printed with two REAL format specifications, representing the real and imaginary parts in that order. It is good practice to use additional format specifiers to print the values in parentheses, or in the '' form. Both forms are illustrated in Figure 27.
PROGRAM COMPLX COMPLEX A,B,C READ(5,100)A,B C = A*B WRITE(6,200)A,B,C 100 FORMAT(2F10.3) 200 FORMAT(1HO,' A = (',F10.3,',',F10.3,')'/ 1 1HO,' B = (',F10.3,',',F10.3,')'/ 2 1H0,' A*B =',F8.3,' + I',F8.3) STOP ENDResults:A = ( 12.500, 8.400) B = ( 6.500 9.600) C = 0.610 + I 174.600
Figure 27: Complex numbers exampleAn operation with two COMPLEX operands always gives a COMPLEX result. In mixed mode expressions, COMPLEX values may be used with REAL or INTEGER, but not with DOUBLE PRECISION values. The REAL or INTEGER value is converted to a COMPLEX value with an imaginary part of zero.
COMPLEX arguments may be used in generic functions such as ABS, EXP, LOG, SQRT, SIN and COS to obtain a COMPLEX value. The following functions are provided for use with COMPLEX values. (C, I, R and D represent COMPLEX, INTEGER, REAL and DOUBLE precision arguments respectively.)
Name Type Definition AIMAG(C) REAL Imaginary part CMPLX(IRD1,IRD2) COMPLEX Complex number: (IRD1,IRD2) CONJG(C) COMPLEX Complex conjugate REAL(C) REAL Real part Figure 28: Some functions used with COMPLEX values
First page
-12- Other Fortran 77 features
- 12.1 EQUIVALENCE
- 12.2 COMMON
- 12.3 BLOCKDATA
- 12.4 Other obsolescent features
- 12.4.1 Arithmetic IF
- 12.4.2 Computed GOTO
12.1 EQUIVALENCE
An EQUIVALENCE statement is used to specify the sharing of the same storage units by two or more variables or arrays.Example:
PROGRAM EQUIV COMPLEX*16 CMPLX(2) REAL*8 TAMPON(4) CHARACTER*8 STR CHARACTER*1 TC(8) EQUIVALENCE (TAMPON(1), CMPLX(1)) EQUIVALENCE (STR, TC(1)) STR = 'ABCDEFGH' DO 10 I=1,4 TAMPON(I)=I 10 CONTINUE PRINT *, 'TC(3)=', TC(3), ' TC(4)=', TC(4) PRINT *, 'CMPLX(1)=', CMPLX(1), ' CMPLX(2)=', CMPLX(2) ENDResult:TC(3)=C TC(4)=D CMPLX(1)=(1.0,2.0) CMPLX(2)=(3.0,4.0)
- The STR CHARACTER*8 variable shares the same memory
storage as the CHARACTER*8 array TC.
- The 4 elements of the REAL*8 array TAMPON are equivalenced with the 2 elements of the COMPLEX*16 array CMPLX. The real and imaginary parts of CMPLX(1) share the same memory storage as TAMPON(1) and TAMPON(2).
< CMPLX(1) >< CMPLX(2) > Memoy ...==><===========><===========><==========><===========><==... < TAMPON(1) >< TAMPON(2) >< TAMPON(3)>< TAMPON(4) > Figure 29: equivalenced objectsNote: if the equivalenced objects have differing type, no conversion nor mathematical equivalence is done.
12.2 COMMON
The COMMON statement specifies blocks of physical storage, called common blocks that may be accessed by any of the functions or subroutines of a program. Thus, the COMMON provides a global facility based on storage association.The common blocks may be named and are called named common blocks, or may be unnamed and are called blank common.
Example:
COMMON X1, TAB(10) COMMON/BLOC1/ A, B, MAT(10, 15) REAL B(50)Note: an EQUIVALENCE statement must not cause storage association of two common blocks and an EQUIVALENCE statement association must not cause a common block storage sequence to be extended. For example, the following is not permitted:
EQUIVALENCE (TAB(1), MAT(1,1)) EQUIVALENCE (TAB(1), B(1))12.3 BLOCKDATA
A block data program unit is used to provide values for data objects in named common blocks.
Example:
BLOCK DATA INIT COMMON/BLOC1/ A, B, MAT(10, 15) DATA A /0./, B /3.14/, MAT /150 * 0.0/ END12.4 Other obsolescent features
These features should not any more be used as they will get out of the future Fortran standards. We document them for those who need to migrate their Fortran 77 old programs to Fortran 90 and above.12.4.1 Arithmetic IF
Its' a conditional GOTO. According to the value of an INTEGER expression, this statement allows branching to one of three specified labels.Example:
IF(I+2) 10, 20 ,30 . . . 10 . . . . . . 20 . . . . . . 30 . . . . . .If the value of the integer expression (here I+2) is :- positve: execution continue at the statement with label 30,
- null: execution continue at the statement with label 20,
- negative: execution continue at the statement with label 10.
Note: in Fortran 90 this feature should be replaced by the IF construct (see IF...THEN...ELSEIF...ENDIF statements) or the CASE construct (see select case statement).
12.4.2 Computed GOTO
Its' a conditional GOTO. According to the value of an INTEGER expression, this statement allows branching to one of a list of specified labels.Example:
IF(10, 20 ,30, 40), I+2 . . . 10 . . . . . . 20 . . . . . . 30 . . . . . . 40 . . . . . .If the value of the integer expression (here I+2) is :- 1: execution continue at the statement with label 10,
- 2: execution continue at the statement with label 20,
- 3: execution continue at the statement with label 30,
- 3: execution continue at the statement with label 40.
Note: in Fortran 90 this feature should be replaced by the CASE construct (see select case statement).
First page
-13- Writing and testing programs
This chapter contains some general principles which should be observed in writing and testing programs. An attempt has been made to demonstrate them in the examples in the previous chapters.
- Plan your programs
Do not attempt to write the Fortran code straight away. Write an outline as shown in the examples, showing the main steps in sequence. Use indentation to indicate the logical structure.
- Develop in stages
The steps defined in 1. can initially be quite broad and general. Revise your outline as often as required, breaking down each main step into a sequence of simpler steps. Repeat this process until your steps correspond to Fortran statements.
- Define variables and arrays
While developing your program as above, think about the main variables and arrays you will require to represent the information. Choose names which suggest their usage and write down each name with its type and dimensions (if an array) and a note of what it represents.
Always use IMPLICIT NONE statement to force explicit typing of variables and arrays.
- Modularise
Use functions and subroutines not only to avoid repetitive coding but, more importantly, to keep your program simple and its structure clear by putting the details of clearly defined computations in separate units.
Don't hesitate to use available scientific libraries as NAG, IMSL, LAPACK, ... which saves you development time and offers you a best optimisation and reliability.
- Provide for exceptions
Your program should be designed to cope with invalid data by printing informative error messages rather than simply failing, e.g. due to attempted division by zero. This applies especially if your program will be used by others.
- Clarity
When writing your Fortran code, use indentation to clarify its logical structure and include explanatory comments freely.
- Testing
Once you have eliminated the syntax errors from your program and subroutines, try running them using suitable test data. Calculate what the results should be, and check that the actual results correspond. If they do not, you will have to revise some of the steps above to correct the errors in your logic. To determine the cause of an error, you may have to insert extra WRITE statements to print out the values of variables etc. at various stages.
Many debugging options or tools can help you finding errors. For the arrays, the "bound checking" options are very usefull to detect errors at execution.
Your test data should be designed to test the various logical paths through your program. For example, to test the quadratic roots program of Figure 7 on page 16, you should use data designed to obtain two, one and no real roots, as well as a value of zero for a.
- Optimizing
Try to use optimisation compiler options to get an improved version of your exectable program. As a second step in this way, you can use profiling tools do detect the more important parts of your program (in term of CPU time) and try to improve them by hand...
First page -14- The standard ASCII table
Character decimal hexa. octal Character decimal hexa. octal C-@ (NUL) 0 0x00 000 espace 32 0x20 040 C-a (SOH) 1 0x01 001 ! 33 0x21 041 C-b (STX) 2 0x02 002 " 34 0x22 042 C-c (ETX) 3 0x03 003 # 35 0x23 043 C-d (EOT) 4 0x04 004 $ 36 0x24 044 C-e (ENQ) 5 0x05 005 % 37 0x25 045 C-f (ACK) 6 0x06 006 & 38 0x26 046 C-g (BEL) 7 0x07 007 ' 39 0x27 047 C-h (BS) 8 0x08 010 ( 40 0x28 050 C-i (HT) 9 0x09 011 ) 41 0x29 051 C-j (LF) 10 0x0a 012 * 42 0x2a 052 C-k (VT) 11 0x0b 013 + 43 0x2b 053 C-l (FF) 12 0x0c 014 , 44 0x2c 054 C-m (CR) 13 0x0d 015 - 45 0x2d 055 C-n (SO) 14 0x0e 016 . 46 0x2e 056 C-o (SI) 15 0x0f 017 / 47 0x2f 057 C-p (DLE) 16 0x10 020 0 48 0x30 060 C-q (DC1) 17 0x11 021 1 49 0x31 061 C-r (DC2) 18 0x12 022 2 50 0x32 062 C-s (DC3) 19 0x13 023 3 51 0x33 063 C-t (DC4) 20 0x14 024 4 52 0x34 064 C-u (NAK) 21 0x15 025 5 53 0x35 065 C-v (SYN) 22 0x16 026 6 54 0x36 066 C-w (ETB) 23 0x17 027 7 55 0x37 067 C-x (CAN) 24 0x18 030 8 56 0x38 070 C-y (EM) 25 0x19 031 9 57 0x39 071 C-z (SUB) 26 0x1a 032 : 58 0x3a 072 C-[ (ESC) 27 0x1b 033 ; 59 0x3b 073 C-\ (FS) 28 0x1c 034 < 60 0x3c 074 C-] (GS) 29 0x1d 035 = 61 0x3d 075 C-$ (RS) 30 0x1e 036 > 62 0x3e 076 C-_ (US) 31 0x1f 037 ? 63 0x3f 077 Character decimal hexa. octal Character decimal hexa. octal @ 64 0x40 100 ` 96 0x60 140 A 65 0x41 101 a 97 0x61 141 B 66 0x42 102 b 98 0x62 142 C 67 0x43 103 c 99 0x63 143 D 68 0x44 104 d 100 0x64 144 E 69 0x45 105 e 101 0x65 145 F 70 0x46 106 f 102 0x66 146 G 71 0x47 107 g 103 0x67 147 H 72 0x48 110 h 104 0x68 150 I 73 0x49 111 i 105 0x69 151 J 74 0x4a 112 j 106 0x6a 152 K 75 0x4b 113 k 107 0x6b 153 L 76 0x4c 114 l 108 0x6c 154 M 77 0x4d 115 m 109 0x6d 155 N 78 0x4e 116 n 110 0x6e 156 O 79 0x4f 117 o 111 0x6f 157 P 80 0x50 120 p 112 0x70 160 Q 81 0x51 121 q 113 0x71 161 R 82 0x52 122 r 114 0x72 162 S 83 0x53 123 s 115 0x73 163 T 84 0x54 124 t 116 0x74 164 U 85 0x55 125 u 117 0x75 165 V 86 0x56 126 v 118 0x76 166 W 87 0x57 127 w 119 0x77 167 X 88 0x58 130 x 120 0x78 170 Y 89 0x59 131 y 121 0x79 171 Z 90 0x5a 132 z 122 0x7a 172 [ 91 0x5b 133 { 123 0x7b 173 \ 92 0x5c 134 | 124 0x7c 174 ] 93 0x5d 135 } 125 0x7d 175 ^ 94 0x5e 136 ~ 126 0x7e 176 _ 95 0x5f 137 C-? 127 0x7f 177 First page
-15- Copyright
Copyright: The University of Strathclyde Computer Centre, Glasgow, Scotland.
(webperson@strath.ac.uk)Permission to copy will normally be granted provided that these credits remain intact.
We'd appreciate a request before you use these notes, partly to justify distributing them, but also so we can distribute news of any updates.
These notes were written by John Porter of the University of Strathclyde Computer Centre. They form the basis of the Computer Centre's Fortran 77 course. John can be reached at J.R.Porter@strath.ac.uk.
Original URL : http://www.strath.ac.uk/CC/Courses/fortran.html
IDRIS adaptation by Hervé Delouis and Patrick Corde.IDRIS/CNRS : bâtiment 506 BP167 - 91403 ORSAY Cedex - France
First page