Function definition

From Rosetta Code
Task
Function definition
You are encouraged to solve this task according to the task description, using any language you may know.

A function is a body of code that returns a value.

The value returned may depend on arguments provided to the function.


Task

Write a definition of a function called "multiply" that takes two arguments and returns their product.

(Argument types should be chosen so as not to distract from showing how functions are created and values returned).


Related task



11l

Function definition:

F multiply(a, b)
   R a * b

Lambda function definition:

V multiply = (a, b) -> a * b

360 Assembly

Linkage conventions are: register 1 : the parameter list, register 0 : the return value, and register 14 : the return address.

DEFFUN   CSECT
         USING  DEFFUN,R13
SAVEAREA B      PROLOG-SAVEAREA(R15)
         DC     17F'0'
PROLOG   STM    R14,R12,12(R13)
         ST     R13,4(R15)
         ST     R15,8(R13)
         LR     R13,R15            set base register
BEGIN    L      R2,=F'13'
         ST     R2,X               X=13
         L      R2,=F'17'
         ST     R2,Y               Y=17
         LA     R1,PARMLIST        R1->PARMLIST
         B      SKIPPARM
PARMLIST DS     0F
         DC     A(X)
         DC     A(Y)
SKIPPARM BAL    R14,MULTPLIC       call MULTPLIC
         ST     R0,Z               Z=MULTPLIC(X,Y)
RETURN   L      R13,4(0,R13)       epilog
         LM     R14,R12,12(R13)
         XR     R15,R15            set return code
         BR     R14                return to caller
*        
MULTPLIC EQU    *                  function MULTPLIC(X,Y)
         L      R2,0(R1)           R2=(A(X),A(Y))
         XR     R4,R4              R4=0     
         L      R5,0(R2)           R5=X    
         L      R6,4(R2)           R6=Y    
         MR     R4,R6              R4R5=R4R5*R6
         LR     R0,R5              R0=X*Y   (R0 return value)
         BR     R14                end function MULTPLIC
*
X        DS     F
Y        DS     F
Z        DS     F
         YREGS  
         END    DEFFUN

6502 Assembly

As with other low-level languages, 6502 assembler has subroutines rather than functions in the strict sense. This implementation of MULTIPLY behaves rather like a function, however: it expects two 'parameters' to be passed in the index registers X and Y and it returns the answer in the accumulator. Note that the 6502 has no MUL instruction, so multiplication is carried out by repeated addition.

MULTIPLY: STX   MULN      ; 6502 has no "acc += xreg" instruction,
          TXA             ; so use a memory address
MULLOOP:  DEY
          CLC             ; remember to clear the carry flag before
          ADC   MULN      ; doing addition or subtraction
          CPY   #$01
          BNE   MULLOOP
          RTS

An alternative implementation that multiplies A by X and checks if A/X is zero.

; https://skilldrick.github.io/easy6502/
; Multiplies A by X

define    memory  1040

          JMP MAIN
 
MULTIPLY: STA memory   ; memory = A
          BEQ MUL_END  ; A = 0
          TXA          ; A = X
          BEQ MUL_END  ; X = 0 -> A = 0
          LDA memory
          CLC
MUL_LOOP: DEX          ; X -= 1
          BEQ MUL_END  ; X = 0 -> A = A * X
          ADC memory   ; A += memory
          JMP MUL_LOOP
MUL_END:  RTS
 
MAIN:     LDA #50
          LDX #5
          JSR MULTIPLY

68000 Assembly

What values are returned (if any) and where they are returned, will depend on the calling convention used. Code written by a C compiler will typically pass parameters onto the stack and use a "frame pointer" to reference them. For this simple example, the operands will be passed into the function using the registers D0 and D1, and the output will be in D0. A function is called by using JSR foo where foo is a labeled section of code or a 24-bit memory address. Execution will continue along starting at that address, until an RTS is encountered, at which point the return address will be popped off the stack into the program counter.

MOVE.L D0,#$0200
MOVE.L D1,#$0400

JSR doMultiply
;rest of program

JMP $           ;halt

;;;;; somewhere far away from the code above
doMultiply:
MULU D0,D1
RTS


8051 Assembly

Like other assembly languages, 8051 doesn't have functions but instead has symbolic references to code. Function arguments are passed via registers decided on beforehand.

ORG RESET
mov a, #100
mov b, #10
call multiply
; at this point, the result of 100*10 = 1000 = 03e8h is stored in registers a and b
; a = e8 
; b = 03
jmp $

multiply:
mul ab
ret

8086 Assembly

A function is nothing more than a named section of code. A CALL instruction will push the current value of the instruction pointer and then set the instruction pointer to that address. Execution will continue forward until a RET statement is encountered, at which point the top of the stack is popped into the instruction pointer register. Note that the RET statement assumes that the top of the stack contains the actual return address, even though in reality this may not be the case. There is no validation that the return address is correct! This is why it's important for the assembly programmer to ensure the stack is balanced at all times, otherwise your program will go running off to who knows where.

It's important to remember that, unlike other languages, execution of assembly code (and this is true for all assembly languages, not just the 8086) is on a purely linear path by default, much like in other "primitive" languages like BASIC, and so there is nothing stopping the instruction pointer from "falling into" subroutines. Often this can be handy if you're trying to code a variation on a function whose only difference is doing a few extra things at the beginning, but it's something you'll need to guard against, either with a return to the operating system or an infinite loop.

start:
mov al, 0x04
mov bl, 0x05
call multiply
;at this point in execution, the AX register contains 0x0900.
;more code goes here, ideally with some sort of guard against "fallthrough" into multiply.

; somewhere far away from start
multiply:
mul bl     ;outputs 0x0014 to ax
ret

AArch64 Assembly

Works with: as version Raspberry Pi 3B version Buster 64 bits
/* ARM assembly AARCH64 Raspberry PI 3B */
/*  program functMul64.s   */

/*******************************************/
/* Constantes file                         */
/*******************************************/
/* for this file see task include a file in language AArch64 assembly*/
.include "../includeConstantesARM64.inc"
 
/***********************/
/* Initialized data */
/***********************/
.data
szRetourLigne: .asciz "\n"
szMessResult:  .asciz "Resultat : @ \n"      // message result
/***********************   
/* No Initialized data */
/***********************/
.bss
sZoneConv:             .skip 24
.text
.global main 
main:
                                // function multiply
    mov x0,8
    mov x1,50
    bl multiply                 // call function
    ldr x1,qAdrsZoneConv
    bl conversion10S            // call function with 2 parameter (x0,x1)
    ldr x0,qAdrszMessResult
    ldr x1,qAdrsZoneConv
    bl strInsertAtCharInc       // insert result at @ character
    bl affichageMess            // display message
 
    mov x0,0                    // return code
 
100:                            // end of  program
    mov x8,EXIT                 // request to exit program
    svc 0                       // perform the system call
qAdrsZoneConv:    .quad sZoneConv
qAdrszMessResult: .quad szMessResult
/******************************************************************/
/*   Function multiply              */ 
/******************************************************************/
/* x0 contains value 1 */
/* x1 contains value 2 */
/* x0 return résult   */
multiply:
    mul x0,x1,x0
    ret               // return function
/********************************************************/
/*        File Include fonctions                        */
/********************************************************/
/* for this file see task include a file in language AArch64 assembly */
.include "../includeARM64.inc"

ACL2

(defun multiply (a b) (* a b))

ActionScript

function multiply(a:Number, b:Number):Number {
    return a * b;
}

Ada

function Multiply (A, B : Float) return Float;

and an implementation of:

function Multiply (A, B : Float) return Float is
begin
   return A * B;
end Multiply;


The Ada 2012 standard provides an even simpler way to define and implement functions:

function Multiply(A, B: Float) return Float is (A * B);


Ada supports generic functions which can take generic formal parameters like the numeric type to use:

generic
   type Number is digits <>;
function Multiply (A, B : Number) return Number;

implemented as:

function Multiply (A, B : Number) return Number is
begin
   return A * B;
end Multiply;

To use this, you need to instantiate the function for each type e.g.

with Multiply;
...
function Multiply_Integer is new Multiply(Number => Integer);
use Multiply_Integer; -- If you must

type My_Integer is Range -100..100;
function Multiply_My_Integer is new Multiply(My_Integer);

Aime

real
multiply(real a, real b)
{
    return a * b;
}

ALGOL 60

begin
    comment Function definition;
    
    integer procedure multiply(a,b);
    integer a,b;
    begin
        multiply:=a*b;
    end;
    
    integer c;
    c:=multiply(2,2);
    outinteger(1,c)
end
Output:
 4

ALGOL 68

PROC multiply = ( LONG REAL a, b ) LONG REAL:
(
  a * b
)

ALGOL W

long real procedure multiply( long real value a, b );
begin
    a * b
end

ALGOL-M

This implementation takes two integers and returns an integer. Note that a function is distinguished from a procedure, which does not return a value.

INTEGER FUNCTION MULTIPLY( A, B );
INTEGER A, B;
BEGIN
    MULTIPLY := A * B;
END;

Amazing Hopper

Hopper has no functions, but they can be declared with macros, which are resolved at compile time. Access to the working stack is global, but "local" variables can be declared in program segments written after the ".locals" clause. Let's look at some examples of declaring "functions".

/* this need data into stack */
#context  Multiplication
    mul
    Return \\
#synon Multiplication   *getproduct

#context-free anothermul
    /* #defn Args(*)  #GENCODE $$$*$$$ #REVLIST=0,mov(#REVLIST);#ENDGEN, */
    Args 'a,b'
    Return ( #(a*b) )\\
#synon anothermul    *getanotherproduct

#include <jambo.h>

#prototype _multiply(_X_,_Y_)
#synon  __multiply   Multiply

Main
    /* "prototipos" of functions and procedures.
       Solves internaly */
    Printnl ( Multiply ( 10, 4 ) )
    Printnl ( __multiply ( 10, 4 ) )
    
    /* definición alternativa 1 */
    Printnl ( Set' 10,4 ', Gosub ' Multiply2 ')
    
    /* aseembler Hopper 1 */
    {10,4} jsub( Multiply3 ), {"\n"} print
    
    /* assembler Hopper 2 */
    {10,4} jsub( Multiply4 ), {"\n"} print
    
    /* context */
    Set '10,4', now get product, and print with newline
    
    /* context-free */
    Set '10,4', and get another product; then print with newline
    
End

.locals  /* Subrutines */

_multiply(a,b)
Return ( Mul(a,b) )

/* Define is macro. Others macros: Function, Procedure:
#defn    Define(_F_,*)       _F_:,#GENCODE $$$*$$$ #REVLIST=0;mov(#REVLIST);#ENDGEN;
#defn    Function(_F_,*)     _F_:,#GENCODE $$$*$$$ #REVLIST=0;mov(#REVLIST);#ENDGEN;
#defn    Procedure(_F_,*)    _F_:,#GENCODE $$$*$$$ #REVLIST=0;mov(#REVLIST);#ENDGEN;
*/
Define 'Multiply2, a,b'
Return ( Mul(a,b) )

Multiply3:
   b=0, mov(b), a=0, mov(a)
   {a,b}mul  /* result into stack */
Return  

Multiply4:
   mul   /* get values from stack, 
            and put result into stack */
back  /* Return */
Output:
40.000000
40.000000
40.000000
40.000000
40.000000
40.000000
40.000000

AmigaE

PROC my_molt(a,b)
-> other statements if needed... here they are not
ENDPROC a*b    -> return value

-> or simplier

PROC molt(a,b) IS a*b

PROC main()
  WriteF('\d\n', my_molt(10,20))
ENDPROC

AntLang

multiply: * /`*' is a normal function
multiply: {x * y}

Explicit definition has the syntax:

{expr-or-def1; expr-or-def2; ..; return-expr}

Inside functions, the variable args contains the sequence of arguments. x, y and z contain the first, second and third argument.

APL

Works with: GNU_APL
⍝⍝ APL2 'tradfn' (traditional function)
⍝⍝ This syntax works in all dialects including GNU APL and Dyalog.
 product  a multiply b
  product  a × b


      ⍝⍝ A 'dfn' or 'lambda' (anonymous function)
      multiply  {×}
Works with: Dyalog_APL
      ⍝⍝ Dyalog dfn (lambda) syntax
      multiply    ×

Works on arrays of any rank (any number of dimensions): atoms, lists, tables, etc.

AppleScript

to multiply(a as number, b as number)
    return a * b
end

A function in AppleScript is called a "handler". It can take one of three different forms, depending on what the scripter finds most convenient. Calls to it must match the form used in the handler definition. Either to or on may be used as the first word in the handler definition. When the script is compiled, the handler label is automatically appended to the end line too if it wasn't written in.

Handler names followed by zero or more parameters within parentheses are called "positional" -- the number and order of the parameters in the caller must match those in the handler definition.

on multiply(a, b)
    return a * b
end multiply

multiply(2, 3)

AppleScript also offers handlers with "prepositional" labeled parameters. These aren't used often because the set of AppleScript-defined prepositions makes it difficult to choose ones that make sense in English.

These prepositions can be used: about, above, against, apart from, around, aside from, at, below, beneath, beside, between, by, for, from, instead of, into, on, onto, out of, over, since, thru, through, and under. Also, of is available, but if used it must be the first parameter.

Example:

on multiplication of a by b
    return a * b
end multiplication

multiplication of 2 by 3 -- Or: (multiplication by 3) of 2, or: 2's (multiplication by 3)

Labeled parameters don't need to be in the same order in the calls as in the handler definition, but of, if used, is regarded as a direct parameter and requires some parenthesis if it's not given first or if the context isn't entirely clear.

For the past few years, handlers with "interleaved" parameters have also been possible. They're a development from AppleScriptObjectiveC and coders can specify their own labels provided these aren't reserved words. Calls to these handlers must reference the handlers' "owners", which are usually represented within the same script by the keyword my. The parameter order is the same in the calls as in the handler definitions:

on multiply:a |by|:b -- 'by' is "barred" here because otherwise it's a reserved word.
    return a * b
end multiply:|by|:

my multiply:2 |by|:3

Argile

use std
.: multiply <real a, real b> :. -> real {a * b}

with a macro and a variable number of parameters:

use std
=: multiply <real a> [<real b>...] := -> real {Cgen a (@@1 (Cgen " * " b))}

ARM Assembly

Works with: as version Raspberry Pi
/* ARM assembly Raspberry PI  */
/*  program functMul.s   */
/* Constantes    */
.equ STDOUT, 1
.equ WRITE,  4
.equ EXIT,   1

/***********************/
/* Initialized data */
/***********************/
.data
szRetourLigne: .asciz "\n"
szMessResult:  .ascii "Resultat : "      @ message result
sMessValeur:   .fill 12, 1, ' '
                   .asciz "\n"
/***********************   
/* No Initialized data */
/***********************/
.bss

.text
.global main 
main:
    push {fp,lr}    /* save  2 registers */

    @ function multiply
	mov r0,#8
	mov r1,#50
	bl multiply             @ call function
    ldr r1,iAdrsMessValeur                
    bl conversion10S       @ call function with 2 parameter (r0,r1)
    ldr r0,iAdrszMessResult
    bl affichageMess            @ display message

    mov r0, #0                  @ return code

100: /* end of  program */
    mov r7, #EXIT              @ request to exit program
    swi 0                       @ perform the system call
iAdrsMessValeur: .int sMessValeur	
iAdrszMessResult: .int szMessResult
/******************************************************************/
/*   Function multiply              */ 
/******************************************************************/
/* r0 contains value 1 */
/* r1 contains value 2 */
/* r0 return résult   */
multiply:
    mul r0,r1,r0
    bx lr	        /* return function */	
	
/******************************************************************/
/*     display text with size calculation                         */ 
/******************************************************************/
/* r0 contains the address of the message */
affichageMess:
    push {fp,lr}    			/* save  registres */ 
    push {r0,r1,r2,r7}    		/* save others registers */
    mov r2,#0   				/* counter length */
1:      	/* loop length calculation */
    ldrb r1,[r0,r2]  			/* read octet start position + index */
    cmp r1,#0       			/* if 0 its over */
    addne r2,r2,#1   			/* else add 1 in the length */
    bne 1b          			/* and loop */
                                /* so here r2 contains the length of the message */
    mov r1,r0        			/* address message in r1 */
    mov r0,#STDOUT      		/* code to write to the standard output Linux */
    mov r7, #WRITE             /* code call system "write" */
    swi #0                      /* call systeme */
    pop {r0,r1,r2,r7}     		/* restaur others registers */
    pop {fp,lr}    				/* restaur des  2 registres */ 
    bx lr	        			/* return  */


/***************************************************/
/*   conversion register in string décimal signed  */
/***************************************************/
/* r0 contains the register   */
/* r1 contains address of conversion area */
conversion10S:
    push {fp,lr}    /* save registers frame and return */
    push {r0-r5}   /* save other registers  */
    mov r2,r1       /* early storage area */
    mov r5,#'+'     /* default sign is + */
    cmp r0,#0       /* négatif number ? */
    movlt r5,#'-'     /* yes sign is - */
    mvnlt r0,r0       /* and inverse in positive value */
    addlt r0,#1
    mov r4,#10   /* area length */
1: /* conversion loop */
    bl divisionpar10 /* division  */
    add r1,#48        /* add 48 at remainder for conversion ascii */	
    strb r1,[r2,r4]  /* store byte area r5 + position r4 */
    sub r4,r4,#1      /* previous position */
    cmp r0,#0     
    bne 1b	       /* loop if quotient not equal zéro */
    strb r5,[r2,r4]  /* store sign at current position  */
    subs r4,r4,#1   /* previous position */
    blt  100f         /* if r4 < 0  end  */
    /* else complete area with space */
    mov r3,#' '   /* character space */	
2:
    strb r3,[r2,r4]  /* store  byte  */
    subs r4,r4,#1   /* previous position */
    bge 2b        /* loop if r4 greather or equal zero */
100:  /*  standard end of function  */
    pop {r0-r5}   /*restaur others registers */
    pop {fp,lr}   /* restaur des  2 registers frame et return  */
    bx lr   


/***************************************************/
/*   division par 10   signé                       */
/* Thanks to http://thinkingeek.com/arm-assembler-raspberry-pi/*  
/* and   http://www.hackersdelight.org/            */
/***************************************************/
/* r0 contient le dividende   */
/* r0 retourne le quotient */	
/* r1 retourne le reste  */
divisionpar10:	
  /* r0 contains the argument to be divided by 10 */
   push {r2-r4}   /* save autres registres  */
   mov r4,r0 
   ldr r3, .Ls_magic_number_10 /* r1 <- magic_number */
   smull r1, r2, r3, r0   /* r1 <- Lower32Bits(r1*r0). r2 <- Upper32Bits(r1*r0) */
   mov r2, r2, ASR #2     /* r2 <- r2 >> 2 */
   mov r1, r0, LSR #31    /* r1 <- r0 >> 31 */
   add r0, r2, r1         /* r0 <- r2 + r1 */
   add r2,r0,r0, lsl #2   /* r2 <- r0 * 5 */
   sub r1,r4,r2, lsl #1   /* r1 <- r4 - (r2 * 2)  = r4 - (r0 * 10) */
   pop {r2-r4}
   bx lr                  /* leave function */
   .align 4
.Ls_magic_number_10: .word 0x66666667

ArnoldC

LISTEN TO ME VERY CAREFULLY multiply
I NEED YOUR CLOTHES YOUR BOOTS AND YOUR MOTORCYCLE a
I NEED YOUR CLOTHES YOUR BOOTS AND YOUR MOTORCYCLE b
GIVE THESE PEOPLE AIR
HEY CHRISTMAS TREE product
YOU SET US UP @I LIED
GET TO THE CHOPPER product
HERE IS MY INVITATION a
YOU'RE FIRED b
ENOUGH TALK
I'LL BE BACK product
HASTA LA VISTA, BABY

Arturo

multiply: $[x,y][x*y]

print multiply 3 7

multiply2: function [x,y][
	return x*y
]

print multiply2 3 7
Output:
21
21

AutoHotkey

MsgBox % multiply(10,2)

multiply(multiplicand, multiplier) {
  Return (multiplicand * multiplier)
}

AutoIt

#AutoIt Version: 3.2.10.0
$I=11
$J=12
MsgBox(0,"Multiply", $I &" * "& $J &" = " & product($I,$J))
Func product($a,$b)
   Return $a * $b
EndFunc

AWK

function multiply(a, b)
{
  return a*b
}
BEGIN {
  print multiply(5, 6)
}

Axe

Lbl MULT
r₁*r₂
Return

BASIC

ANSI BASIC

In ANSI BASIC, functions can be defined as either formulas or multi-line external or internal subroutines. External functions are independent program units that can be called from within the program. Internal functions are considered part of the program unit they are contained in and can only be called from within that unit. External functions do not share any information with other program units and exchange information through parameters and returned values. Internal functions share everything with their surrounding program unit except for their parameters. Internal functions do not have local variables.

Works with: Decimal BASIC
100 DEF Multiply(A, B) = A * B
110 DECLARE FUNCTION MultiplyI
120 DECLARE EXTERNAL FUNCTION MultiplyE
130 PRINT Multiply(3, 1.23456)
140 PRINT MultiplyI(3, 1.23456)
150 PRINT MultiplyE(3, 1.23456)
160 FUNCTION MultiplyI(X, Y)
170    LET MultiplyI = X * Y
180 END FUNCTION
190 END
200 EXTERNAL FUNCTION MultiplyE(A, B)
210 LET MultiplyE = A * B
220 END FUNCTION
Output:
 3.70368 
 3.70368 
 3.70368 

Applesoft BASIC

Applesoft BASIC functions are unary meaning they only take one argument. As the task asks for a multiply function which takes two arguments this poses a problem. To get around this, the multiply function MU takes one argument as the offset into an array of parameters.

Function names in Applesoft BASIC can be longer than two characters but only the first two characters are significant. Function names cannot contain any keywords.

10  DEF  FN MULTIPLY(P) =  P(P) * P(P+1)
20  P(1) = 611 : P(2) = 78 : PRINT  FN MULTIPLY(1)
47658

BASIC256

function multiply(a, b)
    return a * b
end function

BBC BASIC

BBC BASIC supports both single-line and multi-line function definitions. Note that the function name must begin with FN.

Single-line function:

PRINT FNmultiply(6,7)
END

DEF FNmultiply(a,b) = a * b

Multiline function:

DEF FNmultiply(a,b)
LOCAL c
c = a * b
= c

Chipmunk Basic

10 rem Function definition

20 rem ** 1. Function defined as formula. An obsolete way - does not work properly with integer formal parameters (e.g. x%).
30 def fnmultiply(a, b) = a * b

40 rem ** Call the functions
50 print multiply(3,1.23456)
60 print fn multiply(3,1.23456)
70 end

200 rem ** 2. Function defined as subroutine returning a value 
210 sub multiply(a,b)
220   multiply = a*b
230 end sub
Output:
3.70368
3.70368

Commodore BASIC

In Commodore BASIC function definition can consist of any mathematical operation other functions or commands which result in a numeric expression. The definition is limited to single statement, and it accepts only a single argument. When using the function, keyword fn must precede the function name, which itself must be uniquely distinguishable by its first two characters.

10 DEF FN MULT(X) = X*Y
20 Y = 4 : REM VALUE OF SECOND ARGUMENT MUST BE ASSIGNED SEPARATELY
30 PRINT FN MULT(3)

Creative Basic

DECLARE Multiply(N1:INT,N2:INT)

DEF A,B:INT

A=2:B=2

OPENCONSOLE

PRINT Multiply(A,B)

PRINT:PRINT"Press any key to close."

DO:UNTIL INKEY$<>""

CLOSECONSOLE

END

SUB Multiply(N1:INT,N2:INT)

     DEF Product:INT

     Product=N1*N2

RETURN Product

'Can also be written with no code in the subroutine and just RETURN N1*N2.

FreeBASIC

' FB 1.05.0 Win64

Function multiply(d1 As Double, d2 As Double) As Double
  Return d1 * d2
End Function

This function could either be used for all numeric types (as they are implicitly convertible to Double) or could be overloaded to deal with each such type (there are 12 of them).

Alternatively, one could write a macro though this wouldn't be type-safe:

#Define multiply(d1, d2) (d1) * (d2)

Gambas

Click this link to run this code

Public Sub Main()

Print Multiply(56, 4.66)

End

Public Sub Multiply(f1 As Float, f2 As Float) As Float

Return f1 * f2

End

Output:

260.96

GW-BASIC

Works with: BASICA
10 DEF FNMULT(X,Y)=X*Y
20 PRINT FNMULT(5,6)
39 END

IS-BASIC

100 DEF MULTIPLY(A,B)=A*B

IWBASIC

'1. Not Object Oriented Program

DECLARE Multiply(N1:INT,N2:INT),INT
 
DEF A,B:INT
 
A=2:B=2
 
OPENCONSOLE
 
PRINT Multiply(A,B)

PRINT
 
'When compiled as a console only program, a press any key to continue is automatic. 
CLOSECONSOLE
 
END
 
SUB Multiply(N1:INT,N2:INT),INT
 
     DEF Product:INT
 
     Product=N1*N2
 
RETURN Product
ENDSUB

'Can also be written with no code in the subroutine and just RETURN N1*N2.

----

'2. Not Object Oriented Program Using A Macro

$MACRO Multiply (N1,N2) (N1*N2)

DEF A,B:INT

A=5:B=5

OPENCONSOLE

PRINT Multiply (A,B)

PRINT

'When compiled as a console only program, a press any key to continue is automatic.
CLOSECONSOLE

END

----

'3. In An Object Oriented Program

CLASS Associate
'functions/methods
DECLARE Associate:'object constructor
DECLARE _Associate:'object destructor
'***Multiply declared***
DECLARE Multiply(UnitsSold:UINT),UINT
'members
DEF m_Price:UINT
DEF m_UnitsSold:UINT
DEF m_SalesTotal:UINT
ENDCLASS

DEF Emp:Associate

m_UnitsSold=10

Ass.Multiply(m_UnitsSold)

OPENCONSOLE

PRINT"Sales total: ",:PRINT"$"+LTRIM$(STR$(Emp.m_SalesTotal))

PRINT 

CLOSECONSOLE 

END

'm_price is set in constructor
SUB Associate::Multiply(UnitsSold:UINT),UINT
     m_SalesTotal=m_Price*UnitsSold
     RETURN m_SalesTotal
ENDSUB

SUB Associate::Associate()
     m_Price=10
ENDSUB

SUB Associate::_Associate()
'Nothing to cleanup
ENDSUB

Liberty BASIC

Works with: Just BASIC
'     define & call a function

print multiply( 3, 1.23456)

wait

function multiply( m1, m2)
    multiply =m1 *m2
end function

end

Locomotive Basic

10 DEF FNmultiply(x,y)=x*y
20 PRINT FNmultiply(2,PI)

Function names are always preceded by "FN" in Locomotive BASIC. Also, PI is predefined by the interpreter as 3.14159265.

OxygenBasic

'SHORT FORMS:
float multiply(float a,b) = a * b
float multiply(float a,b) { return a * b}

'BASIC FORM:
function multiply(float a, float b) as float
  return a * b
end function

'BASIC LEGACY FORM:
function multiply(byval a as float, byval b as float) as float
  function = a * b
end function

'TEST:
print multiply(pi,2) '6.28...

PureBasic

Procedure multiply(a,b)
  ProcedureReturn a*b
EndProcedure

QBasic

'This function could either be used for all numeric types 
'(as they are implicitly convertible to Double)
FUNCTION multiply# (a AS DOUBLE, b AS DOUBLE)
    multiply = a * b
END FUNCTION
'
' Alternatively, it can be expressed in abbreviated form :
'
DEF FNmultiply# (a AS DOUBLE, b AS DOUBLE) = a * b

PRINT multiply(3, 1.23456)
PRINT FNmultiply#(3, 1.23456)
Output:
 3.703680038452148

QuickBASIC

Works with: QBasic
DECLARE FUNCTION multiply% (a AS INTEGER, b AS INTEGER)

FUNCTION multiply% (a AS INTEGER, b AS INTEGER)
    multiply = a * b
END FUNCTION

REALbasic

Function Multiply(a As Integer, b As Integer) As Integer
  Return a * b
End Function

S-BASIC

S-BASIC is unusual in that the function return value is assigned to the END statement that terminates the function.

function multiply(a, b = integer) = integer
end = a * b

rem - exercise the function

print "The product of 9 times 3 is"; multiply(9, 3)

end
Output:
The product of 9 times 3 is 27

SmallBASIC

func multiply(a, b)
  return a * b
end

print "2 * 3 = "; multiply(2, 3)

If function definition is only one line, def can be used:

def multiply(a, b) = a * b
print "2 * 3 = "; multiply(2, 3)


True BASIC

The FUNCTION and DEF commands are synonymous and can be interchanged.

FUNCTION multiply(a, b)
    LET multiply = a * b
END FUNCTION
!
! Alternatively, it can be expressed in abbreviated form :
!
DEF multiply (a, b) = a * b

END

TI-89 BASIC

multiply(a, b)
Func
  Return a * b
EndFunc

uBasic/4tH

In uBasic you can turn any subroutine into a function with the FUNC() function. It takes one argument, which is the label. Arguments are optional.

Print FUNC(_multiply (23, 65))
End

_multiply Param (2) : Return (a@ * b@)

VBA

Function Multiply(lngMcand As Long, lngMplier As Long) As Long
    Multiply = lngMcand * lngMplier
End Function

To use this function :

Sub Main()
Dim Result As Long
    Result = Multiply(564231, 897)
End Sub

VBScript

function multiply( multiplicand, multiplier )
    multiply = multiplicand * multiplier
end function

Usage:

dim twosquared
twosquared = multiply(2, 2)

Visual Basic

Works with: Visual Basic version VB6 Standard
Function multiply(a As Integer, b As Integer) As Integer
    multiply = a * b
End Function

Call the function

Multiply(6, 111)

Visual Basic .NET

Function Multiply(ByVal a As Integer, ByVal b As Integer) As Integer
    Return a * b
End Function

Call the function

Multiply(1, 1)

Yabasic

sub multiply(a, b)
    return a * b
end sub

Xojo

Function Multiply(ByVal a As Integer, ByVal b As Integer) As Integer
    Return a * b
End Function

Call the function

Dim I As Integer = Multiply(7, 6)

ZX Spectrum Basic

On the ZX Spectrum, function names are limited to one letter. Note that the function becomes effective as soon as it is entered into the program, and does not need to be run

10 PRINT FN m(3,4): REM call our function to produce a value of 12
20 STOP
9950 DEF FN m(a,b)=a*b

Batch File

Windows batch files only have procedures, not functions. Instead, environmental variables can be used as a global shared state.

@ECHO OFF
SET /A result = 0
CALL :multiply 2 3
ECHO %result%
GOTO :eof

:multiply
    SET /A result = %1 * %2
    GOTO :eof

:eof

bc

Works with: GNU bc
define multiply(a, b) { return a*b }

print multiply(2, 3)

BCPL

A function is simply defined as an expression in terms of its arguments.

let multiply(a, b) = a * b

Defining a block of code that executes some statements and then returns a result, is done with a separate valof construct, which can appear wherever an expression may appear, but which is mostly used to define a function containing imperative statements. When used this way, it is equivalent to the functions in most other imperative languages.

let multiply(a, b) = valof
$( // any imperative statements could go here
   resultis a * b
$)

BlitzMax

function multiply:float( a:float, b:float )
    return a*b
end function 

print multiply(3.1416, 1.6180)
Output:
5.08310890

Boo

def multiply(x as int, y as int):
    return x * y

print multiply(3, 2)

Binary Lambda Calculus

In lambda calculus, multiplication on Church numerals is mul = \m \n \f. m (n f) which in BLC is

00 00 00 01 1110 01 110 10

If mul is used several times within an expression E, then they can share the same definition by using (\mul. E)(\m\n\f. m (n f)). For example, the cube function is \n. (\mul. mul n (mul n n)) (\m\n\f. m (n f)) which in BLC is

00 01 00 01 01 10 110 01 01 10 110 110 0000000111100111010

BQN

Tacit definition:

Multiply  ×

With names:

Multiply  {𝕨×𝕩}

Bracmat

multiply=a b.!arg:(?a.?b)&!a*!b;
out$multiply$(123456789.987654321); { writes 121932631112635269 to standard output }

Brat

multiply = { x, y | x * y }

p multiply 3 14  #Prints 42

C

double multiply(double a, double b)
{
   return a * b;
}

Macros

Macros can be defined at the top of a program and the compiler will replace the function calls with the function itself before compiling the program (the source file will not change).

#define MULTIPLY(X, Y) ((X) * (Y))

Parentheses should be added around parameters in the function definition to avoid order of operations errors when someone uses the macro as such:

x = MULTIPLY(x + z, y);

A program with that call would be compiled as if this were coded instead:

x = ((x + z) * (y));

Another advantage of macros is that they work with all types alike. For example, the above macro can be used both to multiply double values (like the function above), and to multiply int values (giving an int, which the function doesn't).

C#

static double multiply(double a, double b)
{
    return a * b;
}

Anonymous function:

Func<double, double, double> multiply = ((a,b) => a*b);

C++

C++ functions basically are the same as in C. Also macros exist, however they are discouraged in C++ in favour of inline functions and function templates.

An inline function differs from the normal function by the keyword inline and the fact that it has to be included in every translation unit which uses it (i.e. it normally is written directly in the header). It allows the compiler to eliminate the function without having the disadvantages of macros (like unintended double evaluation and not respecting scope), because the substitution doesn't happen at source level, but during compilation. An inline version of the above function is:

inline double multiply(double a, double b)
{
   return a*b;
}

If not only doubles, but numbers of arbitrary types are to be multiplied, a function template can be used:

template<typename Number>
  Number multiply(Number a, Number b)
{
   return a*b;
}

Of course, both inline and template may be combined (the inline then has to follow the template<...>), but since templates have to be in the header anyway (while the standard allows them to be compiled separately using the keyword export, almost no compiler implements that), the compiler usually can inline the template even without the keyword.

Since C++20, the template parameters can be inferred using auto:

auto multiply(auto a, auto b)
{
   return a*b;
}

Chapel

proc multiply(a, b)
{
   return a * b;
}

Can require that the two arguments be of the same type.

proc multiply(a : ?t ... 2)
{
   return a(0) * a(1)
}

Will work on any type where the * operator is defined.

ChucK

fun float multiply (float a, float b)
{
    return a * b;
}
// uncomment next line and change values to test
//<<< multiply(16,4) >>>;

Clay

multiply(x,y) = x * y;

Clojure

(defn multiply [x y]
  (* x y))

(multiply 4 5)

Or with multiple arities (in the manner of the actual * function):

(defn multiply
  ([] 1)
  ([x] x)
  ([x y] (* x y))
  ([x y & more] 
    (reduce * (* x y) more)))

(multiply 2 3 4 5)  ; 120

CLU

The following is a function that multiplies two integers and ignores any error conditions (as most examples do).

multiply = proc (a, b: int) returns (int)
    return(a * b)
end multiply

The following is a type-parameterized function that wraps the built-in multiplication operator faithfully, rethrows any exceptions, and works for any type that supports multiplication. It also shows the complete syntax of a function definition (type parameterization, signals, and a where clause).

multiply = proc [T: type] (a, b: T) returns (T) 
           signals (overflow, underflow)
           where T has mul: proctype (T, T) returns (T)
                            signals (overflow, underflow)
    return(a * b) resignal overflow, underflow
end multiply

COBOL

In COBOL, multiply is a reserved word, so the requirements must be relaxed to allow a different function name.

Works with: COBOL-85

The following uses a subprogram:

       IDENTIFICATION DIVISION.
       PROGRAM-ID. myTest.
       DATA DIVISION.
       WORKING-STORAGE SECTION.
       01  x   PICTURE IS 9(3) VALUE IS 3.
       01  y   PICTURE IS 9(3) VALUE IS 2.
       01  z   PICTURE IS 9(9).
       PROCEDURE DIVISION.
           CALL "myMultiply" USING 
               BY CONTENT x, BY CONTENT y, 
               BY REFERENCE z.
           DISPLAY z.
           STOP RUN.
       END PROGRAM myTest.

       IDENTIFICATION DIVISION.
       PROGRAM-ID. myMultiply.
       DATA DIVISION.
       LINKAGE SECTION.
       01  x   PICTURE IS 9(3).
       01  y   PICTURE IS 9(3).
       01  z   PICTURE IS 9(9).
       PROCEDURE DIVISION USING BY REFERENCE x, y, z.
           MULTIPLY x BY y GIVING z.
           EXIT PROGRAM.
       END PROGRAM myMultiply.
Works with: COBOL 2002

This example uses user-defined functions.

       IDENTIFICATION DIVISION.
       PROGRAM-ID. myTest.
       ENVIRONMENT DIVISION.
       CONFIGURATION SECTION.
       REPOSITORY.
           FUNCTION myMultiply.
       DATA DIVISION.
       WORKING-STORAGE SECTION.
       01  x   PICTURE IS 9(3) VALUE IS 3.
       01  y   PICTURE IS 9(3) VALUE IS 2.
       PROCEDURE DIVISION.
           DISPLAY myMultiply(x, y).
           STOP RUN.
       END PROGRAM myTest.

       IDENTIFICATION DIVISION.
       FUNCTION-ID. myMultiply.
       DATA DIVISION.
       LINKAGE SECTION.
       01  x   PICTURE IS 9(3).
       01  y   PICTURE IS 9(3).
       01  z   PICTURE IS 9(9).
       PROCEDURE DIVISION USING x, y RETURNING z.
           MULTIPLY x BY y GIVING z.
           GOBACK.
       END FUNCTION myMultiply.

Coco

As CoffeeScript. In addition, Coco provides some syntactic sugar for accessing the arguments array reminiscent of Perl's @_:

multiply = -> @@0 * @@1

Furthermore, when no parameter list is defined, the first argument is available as it:

double = -> 2 * it

CoffeeScript

multiply = (a, b) -> a * b

ColdFusion

Tag style

<cffunction name="multiply" returntype="numeric">
	<cfargument name="a" type="numeric">
	<cfargument name="b" type="numeric">
	<cfreturn a * b>
</cffunction>

Script style

numeric function multiply(required numeric a, required numeric b){
    return a * b;
}

Common Lisp

Common Lisp has ordinary functions and generic functions.

Ordinary Functions

Ordinary functions operate on the values of argument expressions. Lisp functions terminate by returning one or more values, or by executing a non-local dynamic control transfer, in which case values are not returned.

(defun multiply (a b)
  (* a b))

(multiply 2 3)

User-Defined Compiler Optimization of Functions

In Lisp we can express optimizations of calls to a function using compiler macros. For instance, suppose we know that the multiply function, which may be in another module, simply multiplies numbers together. We can replace a call to multiply by a constant, if the arguments are constant expressions. Like the usual kind of Lisp macro, the compiler macro takes the argument forms as arguments, not the argument values. The special keyword &whole gives the macro access to the entire expression, which is convenient for the unhandled cases, whereby no transformation takes place:

(define-compiler-macro multiply (&whole expr a b)
  (if (and (constantp a) (constantp b))
    (* (eval a) (eval b))
    expr)) ;; no macro recursion if we just return expr; the job is done!

Lisp implementations do not have to honor compiler macros. Usually compilers make use of them, but evaluators do not.

Here is test of the macro using a CLISP interactive session. Note that the multiply function is not actually defined, yet it compiles and executes anyway, which shows that the macro provided the translation something.

$ clisp -q
[1]> (define-compiler-macro multiply (&whole expr a b)
  (if (and (constantp a) (constantp b))
    (* (eval a) (eval b))
    expr))
MULTIPLY
[2]> (defun test1 () (multiply 2 3))
TEST1
[3]> (compile 'test1)
TEST1 ;
NIL ;
NIL
[4]> (disassemble 'test1)

Disassembly of function TEST1
(CONST 0) = 6
[ ... ]
2 byte-code instructions:
0     (CONST 0)                           ; 6
1     (SKIP&RET 1)
NIL
[5]> (test1)
6

Generic Functions

Lisp's generic functions are part of the object system. Generic functions are compiled to ordinary functions, and so are called in the ordinary way. Internally, however, they have the special behavior of dispatching one or more methods based on specializable parameters.

Methods can be defined right inside the DEFGENERIC construct, but usually are written with separate DEFMETHODS.

Also, the DEFGENERIC is optional, since the first DEFMETHOD will define the generic function, but good practice.

;;; terrific example coming

Cowgol

sub multiply(a: int32, b: int32): (rslt: int32) is
    rslt := a * b;
end sub

D

// A function:
int multiply1(int a, int b) {
    return a * b;
}

// Functions like "multiply1" can be evaluated at compile time if
// they are called where a compile-time constant result is asked for:
enum result = multiply1(2, 3); // Evaluated at compile time.
int[multiply1(2, 4)] array;    // Evaluated at compile time.

// A templated function:
T multiply2(T)(T a, T b) {
    return a * b;
}

// Compile-time multiplication can also be done using templates:
enum multiply3(int a, int b) = a * b;

pragma(msg, multiply3!(2, 3)); // Prints "6" during compilation.

void main() {
    import std.stdio;
    writeln("2 * 3 = ", result);
}

Both the compile-time and run-time output:

6
2 * 3 = 6

Dart

main(){
    print(multiply(1,2));
    print(multiply2(1,2));
    print(multiply3(1,2));
}

// the following definitions are equivalent
// arrow syntax without type annotations
multiply(num1, num2) => num1 * num2;

// arrow syntax with type annotations
int multiply2(int num1, int num2) => num1 * num2;

// c style with curly braces
int multiply3(int num1, int num2){
    return num1 * num2;
}

dc

For dc, the functions (called macros) are limited to names from 'a' to 'z' Create a function called 'm'

[*] sm

Use it (lm loads the function in 'm',x executes it, f shows the the stack.)

3 4 lm x f
= 12

Delphi

In addition to what is shown in the section Pascal, the following is possible too:

function multiply(a, b: integer): integer;
begin
  result := a * b;
end;

Diego

begin_funct({number}, multiply)_param({number}, a, b);
    with_funct[]_calc([a]*[b]);
end_funct[];

me_msg()_funct(multiply)_param(1,2);

DM

Functions (called procs) may be derived from proc.

proc/multiply(a, b)
    return a * b

Draco

Draco does not have the equivalent of a return statement. Instead, the last statement in a function must be an expression of the return type of the function.

proc multiply(word a, b) word:
    a * b
corp

Dragon

func multiply(a, b) { 
  return a*b 
}

DWScript

function Multiply(a, b : Integer) : Integer;
begin
   Result := a * b;
end;

Dyalect

func multiply(a, b) {
    a * b
}

Using lambda syntax:

let multiply = (a, b) => a * b

Déjà Vu

multiply a b:
    * a b

E

def multiply(a, b) {
    return a * b
}

(This does not necessarily return a product, but whatever the "multiply" method of a returns. The parameters could be guarded to only accept standard numbers.)

It is also possible to write short anonymous function definitions which do not need explicit returns:

def multiply := fn a, b { a * b }

This definition is identical to the previous except that the function object will not know its own name.

EasyLang

func multiply a b .
   return a * b
.
print multiply 7 5

EchoLisp

(define (multiply a b) (* a b))  multiply ;; (1)
(multiply 1/3 666)  222

;; a function is a lambda definition :
multiply 
      (λ (_a _b) (#* _a _b))

;; The following is the same as (1) :
(define multiply (lambda(a b) (* a b)))
multiply
     (🔒 λ (_a _b) (#* _a _b)) ;; a closure


;; a function may be compiled
(lib 'compile)
(compile 'multiply "-float-verbose")
     
💡 [0]     compiling _🔶_multiply ((#* _a _b))
;; object code (javascript) :
var ref,top = _blocks[_topblock];
/* */return (
/* */(_stack[top] *_stack[1 + top]) 
/* */);

multiply   (λ (_a _b) (#🔶_multiply)) ;; compiled function

Ecstasy

module MultiplyExample {
    static <Value extends Number> Value multiply(Value n1, Value n2) {
        return n1 * n2;
    }

    void run() {
        (Int i1, Int i2) = (7, 3);
        Int i3 = multiply(i1, i2);
        (Double d1, Double d2) = (2.7182818, 3.1415);
        Double d3 = multiply(d1, d2);
        @Inject Console console;
        console.print($"{i1}*{i2}={i3}, {d1}*{d2}={d3}");
    }
}
Output:
7*3=21, 2.7182818*3.1415=8.539482274700001

Efene

multiply = fn (A, B) {
    A * B
}

@public 
run = fn () {
    io.format("~p~n", [multiply(2, 5)])
}

Eiffel

multiply(a, b: INTEGER): INTEGER
	do
		Result := a*b
	end

Ela

multiply x y = x * y

Anonymous function:

\x y -> x * y

Elena

real multiply(real a, real b)
        = a * b;

Anonymous function / closure:

symbol f = (x,y => x * y);

Root closure:

f(x,y){ ^ x * y }

Elixir

defmodule RosettaCode do
  def multiply(x,y) do
    x * y
  end
  
  def task, do: IO.puts multiply(3,5)
end

RosettaCode.task
Output:
15

Elm

--There are multiple ways to create a function in Elm

--This is a named function
multiply x y = x*y

--This is an anonymous function
\x y -> x*y

Emacs Lisp

(defun multiply (x y)
  (* x y))

A "docstring" can be added as follows. This is shown by the Emacs help system and is good for human users. It has no effect on execution.

(defun multiply (x y)
  "Return the product of X and Y."
  (* x y))

EMal

fun multiply = var by var a, var b 
  return a * b
end
writeLine(multiply(6, 7))
writeLine(multiply("can", 2))
Output:
42
cancan

Erlang

Using case, multiple lines

% Implemented by Arjun Sunel
-module(func_definition).
-export([main/0]).

main() ->
	K=multiply(3,4),
	io :format("~p~n",[K]).
	
multiply(A,B) ->
	case {A,B} of 
		{A, B} -> A * B
	end.
Output:
12
ok

In a single line

-module(func_definition).
-export([main/0]).

main() ->
	K=multiply(3,4),
	io :format("~p~n",[K]).
	
multiply(A,B) -> A * B.

The output is the same.

ERRE

A statement function in ERRE is a single line function definition as in Fortran 77 or BASIC. These are useful in defining functions that can be expressed with a single formula. A statement function should appear in declaration part of the program. The format is simple - just type

FUNCTION f(x,y,z,…) 
   f=formula
END FUNCTION

The main features of function statement are:

1) You can use relational operators, so it's possible to "compact" an IF THEN ELSE statement but not loop statements: you must use a procedure for these.

2) Functions can have their own identifier (integer, string, real,double).

3) It's possible to declare function with no parameter: use FUNCTION f()........

4) Functions always return one value.

5) ERRE for C-64 admits only real with one parameter functions.

FUNCTION MULTIPLY(A,B)
   MULTIPLY=A*B
END FUNCTION

Usage:

 IF MULTIPLY(A,B)>10 THEN ......

or

 S=MULTIPLY(22,11)

Euphoria

function multiply( atom a, atom b )
    return a * b
end function

If you declare the arguments as object then sequence comprehension kicks in:

function multiply( object a, object b )
    return a * b
end function

sequence a = {1,2,3,4}
sequence b = {5,6,7,8}

? multiply( 9, 9 )
? multiply( 3.14159, 3.14159 )
? multiply( a, b )
? multiply( a, 7 )
? multiply( 10.39564, b )
Output:
81
9.869587728
{5,12,21,32}
{7,14,21,28}
{51.9782,62.37384,72.76948,83.16512}

F#

The default will be an integer function but you can specify other types as shown:

let multiply x y = x * y // integer
let fmultiply (x : float) (y : float) = x * y

Factor

: multiply ( a b -- a*b ) * ;

Falcon

function sayHiTo( name )
 > "Hi ", name
end

FALSE

[*]     {anonymous function to multiply the top two items on the stack}
m:      {binding the function to one of the 26 available symbol names}
2 3m;!  {executing the function, yielding 6}

Fantom

class FunctionDefinition
{
  public static Void main () 
  {
    multiply := |Int a, Int b -> Int| { a * b }
    echo ("Multiply 2 and 4: ${multiply(2, 4)}")
  }
}

Fermat

Func Multiply(a, b) = a*b.

Fexl

\multiply=(\x\y * x y)

Or if I'm being cheeky:

\multiply=*

Fish

Functions cannot be named in Fish. However, they can be defined as new stacks that pull a certain number of arguments off the stack that came before. 2[ says pull 2 values off the stack and put them in a new, separate stack. ] says put all remaining values in the current stack onto the top of the stack below (the old stack).

2[*]

Forth

: fmultiply ( F: a b -- F: c )  F* ;
: multiply ( a b -- c )  * ;

Fortran

In FORTRAN I (1957), inline function could be defined at the beginning of the program. Let's note than to specify a floating point real the name of the statement function begins with an X (no type declaration) and to specify this is a function the name ends with a F.

     XMULTF(X,Y)=X*Y

And for interger multiplication:

     MULTF(I,J)=I*J

In FORTRAN IV, FORTRAN 66 or later, define a function:

FUNCTION MULTIPLY(X,Y)
REAL MULTIPLY, X, Y
MULTIPLY = X * Y
END

And for integer multiplication:

FUNCTION MULTINT(X,Y)
INTEGER MULTINT, X, Y
MULTINT = X * Y
END

In Fortran 95 or later, define an elemental function, so that this function can be applied to whole arrays as well as to scalar variables:

module elemFunc
contains
    elemental function multiply(x, y)
        real, intent(in) :: x, y
        real :: multiply
        multiply = x * y
    end function multiply
end module elemFunc
program funcDemo
    use elemFunc
    
    real :: a = 20.0, b = 30.0, c
    real, dimension(5) :: x = (/ 1.0, 2.0, 3.0, 4.0, 5.0 /), y = (/ 32.0, 16.0, 8.0, 4.0, 2.0 /), z
    
    c = multiply(a,b)     ! works with either function definition above
    
    z = multiply(x,y)     ! element-wise invocation only works with elemental function
end program funcDemo

It is worth noting that Fortran can call functions (and subroutines) using named arguments; e.g. we can call multiply in the following way:

c = multiply(y=b, x=a)   ! the same as multiply(a, b)
z = multiply(y=x, x=y)   ! the same as multiply(y, x)

(Because of commutativity property of the multiplication, the difference between multiply(x,y) and multiply(y,x) is not evident)

Also note that the function result can be declared with a different name within the routine:

module elemFunc
contains
    elemental function multiply(x, y) result(z)
        real, intent(in) :: x, y
        real :: z
        z = x * y
    end function multiply
end module elemFunc

Free Pascal

Free Pascal allows everything what Delphi allows. Note, using the special variable “result” requires {$modeSwitch result+}. This is the default in {$mode objFPC} and {$mode Delphi}.

Furthermore, after the assignment to the return variable further statements may follow. To ensure a value is returned immediately and no further following statements are processed, using the built-in exit procedure is possible too in {$mode objFPC}:

function multiply(a, b: integer): integer;
begin
  exit(a * b);
end;

If exit has been redefined in the current scope, its special meaning can be accessed via the fully-qualified identifier system.exit. Note, any enclosing finally frames of try … finally … end are processed first before actually returning from the function. As a consequence of that, exit may not appear within a finally frame.

Frink

This function works correctly with any combination of arbitrarily-large integers, arbitrary-precision floating point numbers, arbitrary-size rational numbers, complex numbers, intervals of real numbers, and even numbers with units of measure (e.g. multiply[1 watt, 1 s] gives an answer with dimensions of energy. Frink tries hard to always Do The Right Thing with math and numerics and units of measure.

multiply[x,y] := x*y

Futhark

let multiply (x: i32, y: i32) : i32 = x * y

FutureBasic

window 1

local fn multiply( a as long, b as long ) as long
end fn = a * b

print fn multiply( 3, 9 )

HandleEvents

The function return type can be optionally placed at the beginning of the line:

window 1

long local fn multiply( a as long, b as long )
end fn = a * b

print fn multiply( 3, 9 )

HandleEvents

Output:

27

GAP

multiply := function(a, b)
    return a*b;
end;

GML

In GML one can not define a function but in Game Maker there is a script resource, which is the equivalent of a function as defined here. Scripts can be exported to or imported from a text file with the following format:

#define multiply
a = argument0
b = argument1
return(a * b)

Gnuplot

multiply(x,y) = x*y

# then for example
print multiply(123,456)

Go

Function return types in Go are statically typed and never depend on argument types.

The return statement can contain an expression of the function return type:

func multiply(a, b float64) float64 {
   return a * b
}

Alternatively, if the return value is named, the return statement does not require an expression:

func multiply(a, b float64) (z float64) {
   z = a * b
   return
}

Golfscript

{*}:multiply;

Groovy

def multiply = { x, y -> x * y }

Test Program:

println "x * y = 20 * 50 = ${multiply 20, 50}"
Output:
x * y = 20 * 50 = 1000

Halon

function multiply( $a, $b )
{
    return $a * $b;
}

Haskell

multiply x y = x * y

Alternatively, with help of auto-currying,

multiply = (*)

You can use lambda-function

multiply = \ x y -> x*y

Haxe

function multiply(x:Float, y:Float):Float{
   return x * y;
}

hexiscript

fun multiply a b
  return a * b
endfun

HicEst

FUNCTION multiply(a, b)
   multiply = a * b
END

HolyC

F64 Multiply(F64 a, F64 b) {
  return a * b;
}

F64 x;
x = Multiply(42, 13.37);
Print("%5.2f\n", x);

Hy

Function definition:

(defn multiply [a b]
  (* a b))

Lambda definition:

(def multiply (fn [a b] (* a b)))

i

concept multiply(a, b) {
	return a*b
}

Icon and Unicon

procedure multiply(a,b)
return a * b
end

IDL

The task description is unclear on what to do when the arguments to the function are non-scalar, so here's multiple versions:

function multiply ,a,b
  return, a* b
end

If "a" and "b" are scalar, this will return a scalar. If they are arrays of the same dimensions, the result is an array of the same dimensions where each element is the product of the corresponding elements in "a" and "b".

Alternatively, there's this possibility:

function multiply ,a,b
  return, product([a, b])
end

This will yield the same result for scalars, but if "a" and "b" are arrays it will return the product of all the elements in both arrays.

Finally, there's this option:

function multiply ,a,b
  return, a # b
end

This will return a scalar if given scalars, if given one- or two-dimensional arrays it will return the matrix-product of these arrays. E.g. if given two three-element one-dimensional arrays (i.e. vectors), this will return a 3x3 matrix.

Inform 6

[ multiply a b;
  return a * b;
];

Inform 7

To decide which number is (A - number) multiplied by (B - number):
	decide on A * B.

Io

multiply := method(a,b,a*b)

J

multiply=: *

Works on conforming arrays of any rank (any number of dimensions, as long as the dimensions of one are a prefix of the dimensions of the other): atoms, lists, tables, etc.

Or, more verbosely (and a bit slower, though the speed difference should be unnoticeable in most contexts):

multiply=: dyad define
  x * y
)

Here we use an explicit definition (where the arguments are named) rather than a tacit version (where the arguments are implied). In explicit J verbs, x is the left argument and y is the right argument.

(Note, by the way, that explicit definitions are a subset of tacit definitions -- when the arguments are explicitly named they are still implied in the larger context containing the definition.)

Java

There are no global functions in Java. The equivalent is to define static methods in a class (here invoked as "Math.multiply(a,b)"). Overloading allows us to define the method for multiple types.

public class Math
{
     public static    int multiply(   int a,    int b) { return a*b; }
     public static double multiply(double a, double b) { return a*b; }
}

Using Java version 8

Since version 8, Java has various built in Functions with their normal mathematical names.

import java.util.function.BiFunction;
import java.util.function.DoubleUnaryOperator;
import java.util.function.Function;
import java.util.function.Predicate;
import java.util.function.UnaryOperator;

public final class FunctionDefinition {

	public static void main(String[] args) {
		System.out.println(size.apply("Rosetta"));
		System.out.println(greeting.apply("Joe"));
		System.out.println(half.applyAsDouble(7));
		System.out.println(lessThanTen.test(15));
		System.out.println(add.apply(2, 3));
	}	
	
	private static Function<String, Integer> size = s -> s.length();
	
	private static UnaryOperator<String> greeting = s -> "Hello " + s;
	
	private static DoubleUnaryOperator half = a -> a / 2;
	
	private static Predicate<Integer> lessThanTen = a -> a < 10;
	
	private static BiFunction<Integer, Integer, Integer> add = (a, b) -> a + b;	

}
Output:
7
Hello Joe
3.5
false
5

JavaScript

ES1-*

Function Declaration

function multiply(a, b) { 
  return a*b; 
}

ES3-*

Function Expression

var multiply = function(a, b) {
    return a * b;
};

Named Function Expression

var multiply = function multiply(a, b) {
    return a * b;
};

Method Definition

var o = {
  multiply: function(a, b) {
    return a * b;
  }
};

ES5-*

Accessors

var o = {
  get foo() {
    return 1;
  }, 
  set bar(value) {
    // do things with value
  }
};


ES6-*

Arrow Function

var multiply = (a, b) => a * b;
var multiply = (a, b) => { return a * b };

// Or, `var` being long deprecated, and currying convenient,
// (particularly in use with the higher-order Array methods)

const multiply = a => 
    b => a * b;

Concise Body Method Definition

var o = {
  multiply(a, b) {
    return a * b;
  }
};

Generator Functions

function * generator() {
  yield 1;
}

Joy

DEFINE multiply == * .

jq

Example of a simple function definition:

def multiply(a; b): a*b;

Example of the definition of an inner function:

# 2 | generate(. * .) will generate 2, 4, 16, 256, ...
def generate(f): def r: ., (f | r); r;

The previous example (generate/1) also illustrates that a function argument can be a function or composition of functions. Here is another example:

def summation(f): reduce .[] as $x (0; . + ($x|f));

summation/1 expects an array as its input and takes a function, f, as its argument. For example, if the input array consists of JSON objects with attributes "h" and "w", then to compute SIGMA (h * w) we could simply write:

summation( .h * .w)

Julia

Works with: Julia version 0.6

General function definition:

function multiply(a::Number, b::Number)
  return a * b
end

Julia also supports `assignment` definition as shorthand:

multiply(a, b) = a * b

And lambda calculus:

multiply = (a, b) -> a * b

Kaya

program test;

// A function definition in Kaya:
Int multiply(Int a, Int b) {
    return a * b;
}

// And calling a function:
Void main() {
    putStrLn(string( multiply(2, 3) ));
}

Klingphix

:multiply * ;

2 3 multiply print   { 6 }

Komodo

You can write functions in let expressions or anonimously:

let prod(a, b) := a*b
let prodAnon := (a, b) -> a*b

assert(prodAnon(2, 2) = ((a, b) -> a*b)(2, 2))
assert(prod(2, 2) = prodAnon(2, 2))

Kotlin

// One-liner
fun multiply(a: Int, b: Int) = a * b

// Proper function definition
fun multiplyProper(a: Int, b: Int): Int {
    return a * b
}

Lambdatalk

{def multiply
 {lambda {:a :b}
  {* :a :b}}}

{multiply 3 4}
-> 12

could be written as a variadic function: 

{def any_multiply
 {lambda {:n}   // thanks to variadicity of *
  {* :n}}}

{any_multiply 1 2 3 4 5 6}
-> 720

Lang

Function decleration

fp.multiply = ($a, $b) -> {
	return parser.op($a * $b)
}

One-line function decleration

fp.multiply = ($a, $b) -> return parser.op($a * $b)

Function decleration by using operator functions

fp.multiply = fn.mul

Function decleration by using combinator functions

Combinator functions can be called partially, fn.argCnt2 is used to force the caller to provide 2 arguments to prevent partially calling fp.multiply

fp.multiply = fn.argCnt2(fn.combA2(fn.mul))

Function decleration with call by pointer

fp.multiply = ($[a], $[b]) -> {
	return parser.op($*a * $*b) # Pointers can be dereferenced by using *
}

langur

val multiply = fn{*}
val multiply = fn a, b: a * b
val multiply = fn(a, b) { a * b }

Lasso

Lasso supports multiple dispatch — signature definitions determine which method will be invoked.

define multiply(a,b) => {
	return #a * #b
}

As this function is so simple it can also be represented like so:

define multiply(a,b) => #a * #b

Using multiple dispatch, different functions will be invoked depending on the functions input.

// Signatures that convert second input to match first input
define multiply(a::integer,b::any) => #a * integer(#b)
define multiply(a::decimal,b::any) => #a * decimal(#b)

// Catch all signature 
define multiply(a::any,b::any) => decimal(#a) * decimal(#b)

Latitude

Latitude methods are defined using curly braces {} and assigned to variables like any other value. Arguments are implicitly named $1, $2, etc.

multiply := { $1 * $2. }.

Calling a method is done either with parentheses or with a colon.

multiply (2, 3).
multiply: 2, 3.

If a method is intended to be used as a first-class value or stored in a data structure, the automatic evaluation behavior of methods can be undesired. In this case, one can wrap a method in a Proc with the proc method. Proc objects can then be later called explicitly with call.

multiply := proc { $1 * $2. }.
multiply call (2, 3).
multiply call: 2, 3.

LDPL

data:
n is number

procedure:
sub multiply
    parameters:
        x is number
        y is number
        result is number
    procedure:
        in result solve x * y
end sub

# call the bare sub-procedure
call multiply with 3 4 n
display n lf

# create a statement for it
create statement "multiply $ by $ in $" executing multiply

multiply 3 by 4 in n
display n lf
Output:
12
12

LFE

(defun mutiply (a b)
  (* a b))

Lily

define multiply(a: Integer, b: Integer): Integer
{
  return a * b
}

Lingo

on multiply (a, b)
  return a * b
end

LiveCode

LiveCode has a built-in method called multiply, so there is an extra y to avoid an error.

function multiplyy n1 n2
    return n1 * n2
end multiplyy

put multiplyy(2,5) -- = 10

to multiply :x :y
  output :x * :y
end

LSE64

multiply  : *
multiply. : *.  # floating point

Lua

function multiply( a, b )
    return a * b
end

Lucid

multiply(x,y) = x * y

M2000 Interpreter

A Module can return value

A module can return value to stack of values. Calling a module we place parent stack to module, so we can read any value.

Module Checkit {
      Module Multiply (a, b) {
            Push a*b
      }
      Multiply 10, 5
      Print Number=50
      
      Module Multiply {
            Push Number*Number
      }
      
      Multiply 10, 5
      Print Number=50
      \\ push before call
      Push 10, 5
      Multiply
      Read A
      Print A=50
      Push 10, 2,3 : Multiply : Multiply: Print Number=60
      Module Multiply {
            If not match("NN") Then Error "I nead two numbers"
            Read a, b
            Push a*b
      }
      Call Multiply 10, 5
      Print Number=50
      \\ now there are two values in stack 20 and 50
      Multiply
}
Call Checkit, 20, 50
Print Number=1000

A Local Function Definition

There are two types of function, the normal and the lambda. If a Function return string then we have to use $ at the end of function name.

Module Checkit {
      \\ functions can shange by using a newer definition
      \\ function Multiply is local, and at the exit of Checkit, erased.
      Function Multiply (a, b) {
            =a*b
      }
      Print Multiply(10, 5)=50
      
      Function Multiply {
            =Number*Number
      }
      
      Print Multiply(10, 5)=50

      Function Multiply {
            If not match("NN") Then Error "I nead two numbers"
            Read a, b
            =a*b
      }
      Print Multiply(10, 5)=50
      Function Multiply {
            Read a as long, b as long
            =a*b
      }
      Z=Multiply(10, 5)
      Print Z=50, Type$(Z)="Long"
      Function Multiply(a as decimal=1, b as decimal=2) {
            =a*b
      }
      D=Multiply(10, 5)
      Print D=50, Type$(D)="Decimal"
      D=Multiply( , 50)
      Print D=50, Type$(D)="Decimal"
      D=Multiply( 50)
      Print D=100, Type$(D)="Decimal"
      \\ by reference plus using type
      Function Multiply(&a as decimal, &b as decimal) {
            =a*b
            a++
            b--
      }
      alfa=10@
      beta=20@
      D=Multiply(&alfa, &beta)
      Print D=200, alfa=11,beta=19, Type$(D)="Decimal"
      \\ Using Match() to identify type of items at the top of stack
      Function MultiplyALot {
            M=Stack 
            While Match("NN") {
                  mul=Number*Number
                  Stack M {
                        Data mul  ' at the bottom
                  }
            }
            =Array(M)
      }
      
      K=MultiplyALot(1,2,3,4,5,6,7,8,9,10)
      N=Each(K)
      While N {
            Print Array(N),     ' we get 2  12   30   56   90
      }
      Print
}
Checkit

A Lambda Function

Lambda function is first citizen. We can push it to stack and make another reading from stack. Lambda can use closures as static variables, some of them are pointers so if we copy a lambda we just copy the pointer. Pointers are containers like pointer to array, inventory and stack. Here we define string lambda function (there is a numeric also)

Module CheckIt {
      A$=Lambda$ N$="Hello There" (x) ->{
            =Mid$(N$, x)
      }
      Print A$(4)="lo There"
      Push A$
}
CheckIt
Read B$
Print B$(1)="Hello There"
Function List$ {
      Dim Base 1,   A$()
      A$()=Array$([])  ' make an array from stack items
      =lambda$ A$() (x) -> {
            =A$(x)
      }
 
}
\\ change definition/closures
B$=List$("Hello", "Rosetta", "Function")
Print B$(1)="Hello"

M4

define(`multiply',`eval($1*$2)')

multiply(2,3)

MAD

MAD supports two types of function declarations. One simply evaluates an expression:

            INTERNAL FUNCTION MULT.(A,B) = A * B

Another allows multiple lines to be executed:

            INTERNAL FUNCTION(A, B)
            ENTRY TO MULT.
            FUNCTION RETURN A * B
            END OF FUNCTION

There are several quirks here. First, the length of any identifier must not be longer than six characters, and the name of a function must end in a period (which does not count towards the length). Therefore, the function is called MULT. instead of multiply.

Second, in a multi-line function it is actually the entry point that is named, and a function may have several separate entry points, which need not be at the beginning of the function. Control is transferred to whichever one is called.

Third, all variables are global to the compilation unit. In both examples above, A and B will be set to the values that are passed in, and they will persist after the function has run. They may be declared elsewhere, or they will be of the default type (the NORMAL MODE).

Make

In makefile, a function may be defined as a rule, with recursive make used to retrieve the returned value.

A=1
B=1

multiply:
   @expr $(A) \* $(B)

Invoking it

make -f mul.mk multiply A=100 B=3
> 300

Using gmake, the define syntax is used to define a new function

Works with: gmake
A=1
B=1

define multiply
   expr $(1) \* $(2)
endef

do:
   @$(call multiply, $(A), $(B))

|gmake -f mul.mk do A=5 B=3

Maple

multiply:= (a, b) -> a * b;

Mathematica / Wolfram Language

There are two ways to define a function in Mathematica.

Defining a function as a transformation rule:

multiply[a_,b_]:=a*b

Defining a pure function:

multiply=#1*#2&

Maxima

f(a, b):= a*b;

MAXScript

fn multiply a b =
(
    a * b
)

Mercury

% Module ceremony elided...
:- func multiply(integer, integer) = integer.
multiply(A, B) = A * B.

Metafont

Metafont has macros, rather than functions; through those the language can be expanded. According to the kind of macro we are going to define, Metafont has different ways of doing it. The one suitable for this task is called primarydef.

primarydef a mult b = a * b enddef;
t := 3 mult 5; show t; end

The primarydef allows to build binary operators with the same priority as *. For a more generic macro, we can use instead

def mult(expr a, b) = (a * b) enddef;
t := mult(2,3);
show t;
end

min

'* is syntax sugar for (*), which is an anonymous function that takes two numbers from the data stack, multiplies them, and leaves the result on the data stack. To give it a name, we can use the : sigil which is syntax sugar for define.

'* :multiply

MiniScript

multiply = function(x,y)
    return x*y
end function

print multiply(6, 7)
Output:
42

MiniZinc

function var int:multiply(a: var int,b: var int) = 
    a*b;

МК-61/52

ИП0 ИП1 * В/О

Function (subprogram) that multiplies two numbers. Parameters in registers Р0 and Р1, the result (return value) in register X. Commands ИП0 and ИП1 cause the contents of the corresponding registers in the stack, the more they multiplied (command *) and then code execution goes to the address from which the call subprogram (command В/О).

Modula-2

PROCEDURE Multiply(a, b: INTEGER): INTEGER;
BEGIN
  RETURN a * b
END Multiply;

Modula-3

PROCEDURE Multiply(a, b: INTEGER): INTEGER =
BEGIN
  RETURN a * b;
END Multiply;

MUMPS

MULTIPLY(A,B);Returns the product of A and B
 QUIT A*B

Nanoquery

def multiply(a, b)
    return a * b
end

Neko

var multiply = function(a, b) {
    a * b
}

$print(multiply(2, 3))

Output: 6

Nemerle

public Multiply (a : int, b : int) : int  // this is either a class or module method
{
    def multiply(a, b) { return a * b }   // this is a local function, can take advantage of type inference
    return multiply(a, b)
}

NESL

function multiply(x, y) = x * y;

The NESL system responds by reporting the type it has inferred for the function:

multiply = fn : (a, a) -> a :: (a in number)

NetRexx

/* NetRexx */
options replace format comments java crossref savelog symbols binary

pi      = 3.14159265358979323846264338327950
radiusY = 10
in2ft   = 12
ft2yds  = 3
in2mm   = 25.4
mm2m    = 1 / 1000
radiusM = multiply(multiply(radiusY, multiply(multiply(ft2yds, in2ft), in2mm)), mm2m)

say "Area of a circle" radiusY "yds radius: " multiply(multiply(radiusY, radiusY), pi).format(3, 3) "sq. yds"
say radiusY "yds =" radiusM.format(3, 3)  "metres"
say "Area of a circle" radiusM.format(3, 3)"m radius:" multiply(multiply(radiusM, radiusM), pi).format(3, 3)"m**2"


/**
 * Multiplication function
 */
method multiply(multiplicand, multiplier) public static returns Rexx

  product = multiplicand * multiplier
  return product
Output:
Area of a circle 10 yds radius:  314.159 sq. yds
10 yds =   9.144 metres
Area of a circle   9.144m radius: 262.677m**2

NewLISP

> (define (my-multiply a b) (* a b))
(lambda (a b) (* a b))
> (my-multiply 2 3)
6

Nial

Using variables

multiply is operation a b {a * b}

Using it

|multiply 2 3
=6

Point free form

mul is *

Using it

|mul 3 4
=12

Nial also allows creation of operators

multiply is op a b {a * b}

Using it.

|2 multiply 3
=6
|multiply 2 3
=6

Since this is an array programming language, any parameters can be arrays too

|mul 3 [1,2]
=3 6
|mul [1,2] [10,20]
=10 40

Nim

Nim has a magic variable, `result`, which can be used as a substitute for `return`. The `result` variable will be returned implicitly.

proc multiply(a, b: int): int =
  result = a * b

Here is the same function but with the use of the `return` keyword.

proc multiply(a, b: int): int =
  return a * b

The last statement in a function implicitly is the result value:

proc multiply(a, b: int): int = a * b

OASYS

method int multiply int x int y {
  return x * y
}

OASYS Assembler

OASYS Assembler requires a prefix and suffix on names to indicate their types (an omitted suffix means a void type).

[&MULTIPLY#,A#,B#],A#<,B#<MUL RF

Oberon-2

Oberon-2 uses procedures, and has a special procedure called a "Function Procedure" used to return a value.

PROCEDURE Multiply(a, b: INTEGER): INTEGER;
 BEGIN 
    RETURN a * b;
 END Multiply;

Objeck

function : Multiply(a : Float, b : Float) ~, Float {
   return a * b;
}

OCaml

let int_multiply x y = x * y
let float_multiply x y = x *. y

Octave

function r = mult(a, b)
  r = a .* b;
endfunction

Oforth

Function #* is already defined : it removes 2 objects from the stack and returns on the stack the product of them.

If necessary, we can create a function with name multiply, but, it will just call *

: multiply  * ;

It is also possible to create a function with declared paramaters. In this case, if we define n parameters, n objects will be removed from the stack and stored into those parameters :

: multiply2(a, b)   a b * ;

A function return value (or values) is always what remains on the stack when the function ends. There is no syntax to define explicitely what is the return value(s) of a function.

Ol

Function creation implemented using keyword 'lambda'. This created anonymous function can be saved into local or global variable for further use.

(lambda (x y)
   (* x y))

Ol has two fully equal definitions of global named function (second one is syntactic sugar for first one). In fact both of them is saving the created lambda in global variable.

(define multiply (lambda (x y) (* x y)))

(define (multiply x y) (* x y))

And only one definition of local named functions (with immediate calculation). This type of definition helps to implement local recursions.

(let multiply ((x n) (y m))
   (* x y))

; example of naive multiplication function implementation using local recursion:
(define (multiply x y)
   (let loop ((y y) (n 0))
      (if (= y 0)
         n
         (loop (- y 1) (+ n x)))))

(print (multiply 7 8))
; ==> 56

OOC

multiply: func (a: Double, b: Double) -> Double {
  a * b 
}

ooRexx

Internal Procedure

SAY multiply(5, 6)
EXIT
multiply:
    PROCEDURE
    PARSE ARG x, y
    RETURN x*y

::Routine Directive

say multiply(5, 6)
::routine multiply  
    use arg x, y
    return x *y

Accomodate large factors

say multiply(123456789,987654321)
say multiply_long(123456789,987654321)
::routine multiply
    use arg x, y
    return x *y
::routine multiply_long
    use arg x, y
    Numeric Digits (length(x)+length(y))
    return x *y
Output:
1.21932631E+17
121932631112635269

OpenEdge/Progress

function multiply returns dec (a as dec , b as dec ):
  return a * b .
end.

Oz

fun {Multiply X Y}
   X * Y
end

Or by exploiting first-class functions:

Multiply = Number.'*'

PARI/GP

multiply(a,b)=a*b;

or

multiply=(a,b)->a*b;

Note that in both cases the ; is part of the definition of the function, not of the function itself: it suppresses the output of the function body, but does not suppress the output of the function when called. To do that, either double the semicolon (which will suppress the output of both) or wrap in braces:

multiply={(a,b)->a*b;}

which will return a function which calculates but does not return the product.

Pascal

see also: Delphi and Free Pascal

function multiply(a, b: real): real;
begin
	multiply := a * b
end;

After a function has been activated, there must have be exactly one assignment to the (implicitly declared) variable bearing the same name as of the function. Many processors do not comply with this specification, though, and allow overwriting the return value multiple times.

PascalABC.NET

function multiply(a,b: real): real := a + b;

Perl

The most basic form:

sub multiply { return $_[0] * $_[1] }

or simply:

sub multiply { $_[0] * $_[1] }

Arguments in Perl subroutines are passed in the @_ array, and they can be accessed directly, first one as $_[0], second one as $_[1], etc. When the above function is called with only one or no arguments then the missing ones have an undefined value which is converted to 0 in multiplication.

This is an example using subroutine prototypes:

sub multiply( $$ )
{
   my ($a, $b) = @_;
   return $a * $b;
}

The above subroutine can only be called with exactly two scalar values (two dollar signs in the signature) but those values may be not numbers or not even defined. The @_ array is unpacked into $a and $b lexical variables, which are used later.

The arguments can be automatically unpacked into lexical variables using the experimental signatures feature (in core as of 5.20):

use experimental 'signatures';
sub multiply ($x, $y) {
    return $x * $y;
}

Phix

Library: Phix/basics
with javascript_semantics
function multiply(atom a, atom b)
    return a*b
end function

Phixmonti

def multiply * enddef

PHL

@Integer multiply(@Integer a, @Integer b) [
	return a * b;
]

PHP

function multiply( $a, $b )
{
    return $a * $b;
}

Picat

multiply(A, B) = A*B.

PicoLisp

(de multiply (A B)
   (* A B) )

Pike

int multiply(int a, int b){
   return a * b;
}

PL/I

PRODUCT: procedure (a, b) returns (float);
   declare (a, b) float;
   return (a*b);
end PRODUCT;

PL/SQL

FUNCTION multiply(p_arg1 NUMBER, p_arg2 NUMBER) RETURN NUMBER 
IS
  v_product NUMBER;
BEGIN
  v_product := p_arg1 * p_arg2;
  RETURN v_product;
END;

Plain English

The Multiply a number by another number routine is already defined in the noodle, so we need to tweak the wording slightly so the compiler doesn't complain about redefinition (or so the definition isn't recursive). Note that the number refers to the parameter a number and the other number refers to the parameter another number.

To multiply a number with another number:
Multiply the number by the other number.

Pop11

define multiply(a, b);
    a * b
enddefine;

PostScript

Inbuilt:

3 4 mul

Function would be:

/multiply{
    /x exch def
    /y exch def
    x y mul =
}def

PowerShell

The most basic variant of function definition would be the kind which uses positional parameters and therefore doesn't need to declare much:

function multiply {
    return $args[0] * $args[1]
}

Also, the return statement can be omitted in many cases in PowerShell, since every value that "drops" out of a function can be used as a "return value":

function multiply {
    $args[0] * $args[1]
}

Furthermore, the function arguments can be stated and named explicitly:

function multiply ($a, $b) {
    return $a * $b
}

There is also an alternative style for declaring parameters. The choice is mostly a matter of personal preference:

function multiply {
    param ($a, $b)
    return $a * $b
}

And the arguments can have an explicit type:

function multiply ([int] $a, [int] $b) {
    return $a * $b
}

Processing

Processing is based on Java, and thus uses a familiar C-style syntax for function definition—as it does for much else. For the sake of argument, this implementation of multiply uses single-precision floats: other numeral types are available.

float multiply(float x, float y)
{
    return x * y;
}

Processing Python mode

Processing Python mode is based on Jython, a fully implemented Python 2 interpreter, and thus uses familiar Python syntax for function definition-as it does for much else.

def multiply(x, y):
    return x * y

Prolog

Prolog, as a logic programming languages, does not have user-supplied functions available. It has only predicates; statements which are "true" or "false". In cases where values have to be "returned" a parameter is passed in that is unified with the result. In the following predicate the parameter "P" (for "Product") is used in this role. The following code will work in any normal Prolog environment (but not in things like Turbo Prolog or Visual Prolog or their ilk):

multiply(A, B, P) :- P is A * B.

This is what it looks like in use:

go :-
  multiply(5, 2, P),
  format("The product is ~d.~n", [P]).

This can be a little bit jarring for those used to languages with implicit return values, but it has its advantages. For example unit testing of such a predicate doesn't require special frameworks to wrap the code:

test_multiply :-
  multiply(5, 2, 10),  % this will pass
  multiply(3, 4, 11).  % this will not pass

Still, the lack of user-defined functions remains an annoyance.

Prolog, however, is a remarkably malleable language and through its term re-writing capabilities the function-style approach could be emulated. The following code relies on the function_expansion pack (separately installed through the packs system) for SWI-Prolog. Similar code could be made in any Prolog implementation, however.

:- use_module(library(function_expansion)).

user:function_expansion(multiply(A, B), P, P is A * B).  % "function" definition

go :-
  format("The product is ~d.~n", [multiply(5, 2)]).

While the function definition is perhaps a bit more involved, the function use is now pretty much the same as any other language people are used to. The "magic" is accomplished by the compiler rewriting the go/0 term into the following code:

go :-
  A is 5*2,
  format('The product is ~d.~n', [A]).

Python

Function definition:

def multiply(a, b):
    return a * b

Lambda function definition:

multiply = lambda a, b: a * b

A callable class definition allows functions and classes to use the same interface:

class Multiply:
    def __init__(self):
        pass
    def __call__(self, a, b):
        return a * b

multiply = Multiply()
print multiply(2, 4)    # prints 8

(No extra functionality is shown in this class definition).

Q

multiply:{[a;b] a*b}

or

multiply:{x*y}

or

multiply:*

Using it

multiply[2;3]
 6

Quack

You have several ways to define a function in Quack. You can do it by the classic way:

fn multiply[ a; b ]
  ^ a * b
end

Using lambda-expressions:

let multiply :- fn { a; b | a * b }

And using partial anonymous functions:

let multiply :- &(*)

Quackery

[ * ] is multiply ( n n --> n )

In the Quackery shell (REPL):

/O> 2 3 multiply
...

Stack: 6

Quackery is a stack language: arguments are assumed to be on the stack when functions are called. This means that we don't need to name the parameters of a function. For this reason, we call functions words, because in code they really are just words written one after the other.

( n n --> n ) is a comment that indicates multiply takes two numbers from the data stack and leaves one number on the data stack afterward. Stack comments are not necessary, but they are good form. They show how words interact with the data stack at a glance.

Words don't have to be named. We could have written the above as:

2 ' [ * ] 3 swap do

By quoting the nest containing * with the ' word, we have prevented it from being executed immediately and placed it on the data stack. Now it can be manipulated like any other nest or data stack object. We can use do to execute the contents of the nest.

R

mult <- function(a,b) a*b

In general:

mult <- function(a,b) {
  a*b
  # or:
  # return(a*b)
}

Racket

A simple function definition that takes 2 arguments.

(define (multiply a b) (* a b))

Using an explicit lambda or λ is completely equivalent:

(define multiply (lambda (a b) (* a b)))
(define multiply (λ (a b) (* a b)))

Note that * is a function value, so the following code also works (although multiply will now be variadic function).

(define multiply *)

Raku

(formerly Perl 6) Without a signature:

sub multiply { return @_[0] * @_[1]; }

The return is optional on the final statement, since the last expression would return its value anyway. The final semicolon in a block is also optional. (Beware that a subroutine without an explicit signature, like this one, magically becomes variadic (rather than nullary) only if @_ or %_ appear in the body.) In fact, we can define the variadic version explicitly, which still works for two arguments:

sub multiply { [*] @_ }

With formal parameters and a return type:

sub multiply (Rat $a, Rat $b --> Rat) { $a * $b }

Same thing:

my Rat sub multiply (Rat $a, Rat $b) { $a * $b }

It is possible to define a function in "lambda" notation and then bind that into a scope, in which case it works like any function:

my &multiply := -> $a, $b { $a * $b };

Another way to write a lambda is with internal placeholder parameters:

my &multiply := { $^a * $^b };

(And, in fact, our original @_ above is just a variadic self-declaring placeholder argument. And the famous Perl "topic", $_, is just a self-declared parameter to a unary block.)

You may also curry both built-in and user-defined operators by supplying a * (known as "whatever") in place of the argument that is not to be curried:

my &multiply := * * *;

This is not terribly readable in this case due to the visual confusion between the whatever star and the multiplication operator, but Perl knows when it's expecting terms instead of infixes, so only the middle star is multiplication. It tends to work out much better with other operators. In particular, you may curry a cascade of methods with only the original invocant missing:

@list.grep( *.substr(0,1).lc.match(/<[0..9 a..f]>/) )

This is equivalent to:

@list.grep( -> $obj { $obj.substr(0,1).lc.match(/<[0..9 a..f]>/) } )

Raven

define multiply use a, b
    a b *

Or optional infix:

define multiply use a, b
    (a * b)

Or skip named vars:

define multiply *

REBOL

REBOL actually already has a function called 'multiply', which is a native compiled function. However, since it's not protected, I can easily override it:

multiply: func [a b][a * b]

Relation

function multiply(a,b)
set result = a*b
end function

Retro

: multiply ( nn-n ) * ;

REXX

exactitudeness

multiply: return arg(1) * arg(2)    /*return the product of the two arguments.*/

cleaner display

Because REXX will return the same precision as the multiplicands, we can do some beautification with the resultant product.

I.E.:             3.0 * 4.00     yields the product:     12.000

This version eliminates the   .000   from the product.

multiply: return arg(1) * arg(2) / 1    /*return with a normalized product of 2 args. */

Ring

func multiply x,y return x*y

RLaB

In RLaB the functions can be built-in (compiled within RLaB, or part of the shared object library that is loaded per request of user), or user (written in RLaB script). Consider an example:

>> class(sin)
function
>> type(sin)
builtin

Functions are a data class on their own, or they can be member of a list (associative array).

1. user function specified from built-in functions, here basic addition

f = function(x, y)
{
  return x + y;
};

>> class(f)
function
>> type(f)
user

2. function can be member of a list (associative array)

somelist = <<>>;
somelist.f = function(x, y)
{
  rval = x + y;
  return rval;
};

3. user function which uses a function that is specified as a member of some list, here we use somelist from above:

g = function(x, y)
{
  global(somelist);
  rval = x * somelist.f(x, 2*y);
  return rval;
};

RPL

≪ * ≫ 'MULT' STO
2 3 MULT
Output:
6

Ruby

def multiply(a, b)
    a * b
end

Ruby 3.0 adds endless method definition:

def multiply(a, b) = a * b

Rust

fn multiply(a: i32, b: i32) -> i32 {
    a * b
}

Sather

class MAIN is
  -- we cannot have "functions" (methods) outside classes
  mult(a, b:FLT):FLT is return a*b; end;

  main is
    #OUT + mult(5.2, 3.4) + "\n";
  end;
end;

Scala

def multiply(a: Int, b: Int) = a * b

Scheme

(define multiply *)

Alternately,

(define (multiply a b)
  (* a b))

Seed7

const func float: multiply (in float: a, in float: b) is
  return a * b;

SenseTalk

put multiply(3,7) as words

to multiply num1, num2
	return num1 * num2
end multiply
Output:
twenty-one

SETL

proc multiply( a, b );
    return a * b;
end proc;

Sidef

func multiply(a, b) {
    a * b;
}

Simula

Simula uses the term procedure for subroutines/methods whether they return a value or not. A procedure that does return a value is declared with a data type (e.g. integer procedure), whereas one that does not is declared simply as procedure. This program defines multiply as an integer procedure and illustrates its use. Note that the second argument provided to Outint gives the width of the integer to be printed.

BEGIN
    INTEGER PROCEDURE multiply(x, y);
    INTEGER x, y;
    BEGIN
        multiply := x * y
    END;
    Outint(multiply(7,8), 2);
    Outimage
END

Slate

define: #multiply -> [| :a :b | a * b].

or using a macro:

define: #multiply -> #* `er.

The block may also be installed as a method like so:

a@(Number traits) multiplyBy: b@(Number traits) [a * b].

or more explicitly (without sugar):

[| :a :b | a * b] asMethod: #multipleBy: on: {Number traits. Number traits}.

Smalltalk

|mul|
mul := [ :a :b | a * b ].

SNOBOL4

          define('multiply(a,b)') :(mul_end)
multiply  multiply = a * b        :(return)
mul_end
* Test
          output = multiply(10.1,12.2)
          output = multiply(10,12)
end
Output:
   123.22
   120

SNUSP

For expediency, the function is adding three values, instead of multiplying two values. Another function, atoi (+48) is called before printing the result.

+1>++2=@\=>+++3=@\==@\=.=#  prints '6'
        |        |   \=itoa=@@@+@+++++#
        \=======!\==!/===?\<#
                     \>+<-/

SPARK

The function definition (multiplies two standard Integer):

package Functions is
   function Multiply (A, B : Integer) return Integer;
   --# pre A * B in Integer; -- See note below
   --# return A * B; -- Implies commutativity on Multiply arguments
end Functions;

Note: how do you ensure then “A * B in Integer” ? Either with a proof prior to Multiply invokation or using another form of Multiply where input A and B would be restricted to a range which ensures the resulting product is always valid. Exemple :

type Input_Type is range 0 .. 10;
type Result_Type is range 0 .. 100;

and had a version of Multiply using these types. On the other hand, if arguments of Multiply are constants, this is provable straight away.

The Multiply's implementation:

package body Functions is
   function Multiply (A, B : Integer) return Integer is
   begin
      return A * B;
   end Multiply;
end Functions;

SPL

Single-line function definition:

multiply(a,b) <= a*b

Multi-line function definition:

multiply(a,b)=
  x = a*b
  <= x
.

SSEM

The SSEM instruction set makes no explicit provision for subroutines, and indeed its storage space is too small for them to be of much use; but something like a subroutine can be created using a modified form of Wheeler jump. In this technique, the jump to the subroutine is accomplished with the return address loaded in the accumulator. The first action by the subroutine is to store this address in a place where it will be found by its own final jump instruction. In principle, therefore, the subroutine can be called multiple times from different points in the program without the calling routine needing to modify it at all (or even to know anything about it beyond where it begins, where it expects to find its parameters, and where it will store its result or results).

In this example, the main routine does nothing at all beyond calling the subroutine and halting after it has returned. The values A and B are passed in the two addresses located immediately before the subroutine begins; their product is returned in the address that formerly stored A. Given that the multiply subroutine begins at address 8, the calling routine looks like this:

01000000000000100000000000000000   0. -2 to c
00100000000000000000000000000000   1. 4 to CI
01111111111111111111111111111111   2. -2
00000000000001110000000000000000   3. Stop
11100000000000000000000000000000   4. 7

or in pseudocode:

          load       &here
          jump       multiply
here:     halt

Implementing multiply on the SSEM requires the use of repeated negation and subtraction. For the sake of example, the values 8 and 7 are provided for A and B.

00010000000000000000000000000000   6. 8
11100000000000000000000000000000   7. 7
11111000000001100000000000000000   8. c to 31
01100000000000100000000000000000   9. -6 to c
01111000000001100000000000000000  10. c to 30
01111000000000100000000000000000  11. -30 to c
01111000000001100000000000000000  12. c to 30
11100000000000100000000000000000  13. -7 to c
11100000000001100000000000000000  14. c to 7
11100000000000100000000000000000  15. -7 to c
00111000000000010000000000000000  16. Sub. 28
11100000000001100000000000000000  17. c to 7
00111000000000010000000000000000  18. Sub. 28
00000000000000110000000000000000  19. Test
00111000000001000000000000000000  20. Add 28 to CI
11111000000000000000000000000000  21. 31 to CI
01100000000000100000000000000000  22. -6 to c
01111000000000010000000000000000  23. Sub. 30
01100000000001100000000000000000  24. c to 6
01100000000000100000000000000000  25. -6 to c
01100000000001100000000000000000  26. c to 6
10111000000000000000000000000000  27. 29 to CI
10000000000000000000000000000000  28. 1
00110000000000000000000000000000  29. 12
00000000000000000000000000000000  30. 0
00000000000000000000000000000000  31. 0

The pseudocode equivalent clarifies how the subroutine works, or how it would work on an architecture that supported load and add:

a:        equals     #8
b:        equals     #7
multiply: store      ret
          load       a
          store      n
loop:     load       b
          sub        #1
          store      b
          sub        #1
          ifNegative done
          load       a
          add        n
          store      a
          jump       loop
done:     jump       *ret
n:        reserve    1 word
ret:      reserve    1 word

Standard ML

val multiply = op *

Equivalently,

fun multiply (x, y) = x * y

Using lambda syntax:

val multiply = fn (x, y) => x * y

Curried form:

fun multiply x y = x * y

Stata

Ado

Stata's macro language does not have functions, but commands. Output is usually saved as a "stored result" (but could also be saved in a global macro variable, in a scalar or matrix, in a dataset or simply printed to the Results window). See program and [1] in Stata documentation.

prog def multiply, return
	args a b
	return sca product=`a'*`b'
end

multiply 77 13
di r(product)

Output

1001

Mata

Mata is the matrix language of Stata. Here is how to define a function

mata
scalar multiply(scalar x, scalar y) {
	return(x*y)
}

multiply(77,13)
end

Output

1001

Swift

func multiply(a: Double, b: Double) -> Double {
   return a * b
}

Tcl

Strictly as described in the task:

proc multiply { arg1 arg2 } {
    return [expr {$arg1 * $arg2}]
}
Works with: Tcl version 8.5

You can also create functions that work directly inside expressions. This is done by creating the command with the correct name (that is, in the tcl::mathfunc namespace):

proc tcl::mathfunc::multiply {arg1 arg2} {
    return [expr {$arg1 * $arg2}]
}

# Demonstrating...
if {multiply(6, 9) == 42} {
    puts "Welcome, Citizens of Golgafrincham from the B-Ark!"
}

Toka

[ ( ab-c ) * ] is multiply

Transd

multiply: (lambda a Double() b Double() (* a b))

TXR

In TXR, there are pattern functions which are predicates that perform pattern matching and variable capture. A call to this type of function call can specify unbound variables. If the function succeeds, it can establish bindings for those variables.

Here is how to make a pattern function that multiplies, and call it. To multiply the numbers, we break out of the pattern language and invoke Lisp evaluation: @(* a b)

@(define multiply (a b out))
@(bind out @(* a b))
@(end)
@(multiply 3 4 result)
$ txr -B multiply.txr
result="12"

In the embedded Lisp dialect, it is possible to write an ordinary function that returns a value:

(defun mult (a b) (* a b))
  (put-line `3 * 4 = @(mult 3 4)`)
$ txr multiply.tl
3 * 4 = 12

UNIX Shell

Note that in the Unix shell, function definitions do not include any argument specifications within the parentheses. Instead arguments to functions are obtained using the positional parameters.

Works with: Bourne Shell
multiply() {
  # There is never anything between the parentheses after the function name
  # Arguments are obtained using the positional parameters $1, and $2
  # The return is given as a parameter to the return command
  return `expr "$1" \* "$2"`    # The backslash is required to suppress interpolation
}
 
# Call the function
multiply 3 4    # The function is invoked in statement context
echo $?        # The dollarhook special variable gives the return value
Works with: Bash

return an exit code

multiply() {
  return $(($1 * $2))
}
 
multiply 5 6
echo $?

echo the result

multiply() {
  echo -n $(($1 * $2))
}
 
echo $(multiply 5 6)

Ursa

# multiply is a built-in in ursa, so the function is called mult instead
def mult (int a, int b)
	return (* a b)
end

Ursala

Functions are declared with an equals sign like constants of any other type. They may be specified by lambda abstraction, with dummy variables in double quotes, or in point-free form, or any combination. The way multiplication is defined depends on the type of numbers being multiplied. For this example, numbers in standard IEEE double precision are assumed, and the multiply function is defined in terms of the system library function, called using the syntax math..mul. This is the definition in point free form,

multiply = math..mul

this is the definition using lambda abstraction

multiply = ("a","b"). math..mul ("a","b")

and this is the definition using pattern matching.

multiply("a","b") = math..mul ("a","b")

V

V uses stack for input arguments and '.' is a word that takes a quote and binds the first word to the sequence of actions supplied in the quote.

[multiply *].

Using it

2 3 multiply
=6

V also allows internal bindings.

[multiply
  [a b] let
  a b *].

V (Vlang)

fn multiply(a f64, b f64) f64 {
  return a * b
}

fn main() {
  print(multiply(5, 6))
}
Output:
30.0

Wart

A straightforward way to say how calls of the form (multiply a b) are translated:

def (multiply a b)
  a*b
(multiply 3 4)
=> 12

Functions can also use keyword args.

(multiply 3 :a 4)  # arg order doesn't matter here, but try subtract instead
=> 12

Finally, we can give parameters better keyword args using aliases:

def (multiply a b|by)
  (* a b)
multiply 3 :by 4
=> 12

WebAssembly

   (func $multipy (param $a i32) (param $b i32) (result i32)
       local.get $a
       local.get $b
       i32.mul
   )

Wren

The following 'multiply' function will work for any type(s) that support the '*' operator. However, it will produce a runtime error otherwise, as demonstrated by the final example.

var multiply = Fn.new { |a, b| a * b }

System.print(multiply.call(3, 7))
System.print(multiply.call("abc", 3))
System.print(multiply.call([1], 5))
System.print(multiply.call(true, false))
Output:
21
abcabcabc
[1, 1, 1, 1, 1]
Bool does not implement '*(_)'.
[./function_definition line 1] in new(_) block argument
[./function_definition line 6] in (script)

X86 Assembly

X86 Assembly doesn't really have functions. Instead, it has labels that are called. Function arguments can be pushed onto the stack prior to calling or passed to the function in registers. The system will usually have some sort of calling conventions to facilitate inter-operation between languages.

Unix

Function definition and calling conventions on a Unix-like system are specified in the book "System V Application Binary Interface: Intel 386 Architecture Processor Supplement" (from SCO at archive.org). These are the conventions used by the C language and also most other languages.

The stack, for two 32-bit integer parameters, is

  • [esp+8] second parameter
  • [esp+4] first parameter
  • [esp] return address

The return value is left in the eax register. ecx and edx are "scratch" registers meaning the called routine doesn't need to preserve their values. (In the code below edx is clobbered.)

The following is Unix-style "as" assembler syntax (including GNU as). The resulting function can be called from C with multiply(123,456).

        .text
        .globl  multiply
        .type   multiply,@function
multiply:
        movl    4(%esp), %eax
        mull    8(%esp)
        ret

The .type directive is important for code which will go into a shared library. You can get away without it for a static link. It ensures the linker knows to dispatch calls from the mainline to the function via a PLT entry. (If omitted the code is copied at runtime into some mainline space. Without a .size directive only 4 bytes will be copied.)

NASM

Works with: NASM
section .text
global _start
 
_multiply_regs:
  mul ebx
  mov eax, ebx
  ret
 
_multiply_stack:
  enter 2,0
  mov eax, [esp+4]
  mov ebx, [esp+8]
  mul ebx
  mov eax, ebx 
  leave
  ret
 
_start:
  mov ax, 6  ;The number to multiply by
  mov ebx, 16 ;base number to multiply.
  call _multiply_regs
  push 6
  push 16
  call _multiply_stack

MASM

However, in MASM we do have function statements due to the preprocessor.

Works with: MASM
multiply proc arg1:dword, arg2:dword
  mov eax, arg1
  mov ebx, arg2
  mul ebx
  mov eax, ebx
  ret
multiply endp

Then to call it.

invoke multiply, 6, 16
;or..
push 16
push 6
call multiply

Return values are usually put into the register EAX. This, of course is not a must it's simply that it's somewhat of a unofficial standard. For example, C/C++ preprocessors/compilers will translate "return value" into "mov eax, value" followed by the return to caller instruction "ret".

XBS

Functions are defined by using the func keyword.

func multiply(a,b){
	send a*b;
}

XLISP

Functions can be defined using either 'classic' Lisp syntax:

(defun multiply (x y)
    (* x y))

or Scheme-style syntax:

(define (multiply x y)
    (* x y))

or, if you prefer, with LAMBDA:

(define multiply
    (lambda (x y) (* x y)))

XPL0

func Multiply(A, B);    \the characters in parentheses are only a comment
int  A, B;              \the arguments are actually declared here, as integers
return A*B;            \the default (undeclared) function type is integer
                        \no need to enclose a single statement in brackets
 
func real FloatMul(A, B); \floating point version
real A, B;              \arguments are declared here as floating point (doubles)
return A*B;

XSLT

Templates are the closest things XSLT has to user defined functions. They can be declared to be called by name and/or to be applied to all nodes in a matching set and given "mode". Both types of template can take named parameters with default values. Templates also have a "context" node used as the base of XPath expressions (kind of like an implied "this" of an object's method).

<xsl:template name="multiply">
  <xsl:param name="a" select="2"/>
  <xsl:param name="b" select="3"/>
  <xsl:value-of select="$a * $b"/>
</xsl:template>

Usage examples.

<xsl:call-template name="multiply">
  <xsl:with-param name="a">4</xsl:with-param>
  <xsl:with-param name="b">5</xsl:with-param>
</xsl:call-template>
 
<xsl:call-template name="multiply"/>    <-- using default parameters of 2 and 3 -->

Available in XSLT 2.0 and later versions.

<xsl:function name="mf:multiply">
    <xsl:param name="a"/>
    <xsl:param name="b"/>
    <xsl:value-of select="$a * $b"/>
</xsl:function>

Usage examples.

{mf:multiply(2,3)}
<xsl:value-of select="mf:multiply(2,3)" />

YAMLScript

!yamlscript/v0

# Main function definition with variable arguments:
defn main(*args):
  say: "multiply($(args.join(', ')))
        -> $multiply(args*)"

# A multi-arity function definition:
defn multiply:
  (): 1
  (x): x
  (x y): x * y
  (x y *more):
    reduce multiply:
      multiply(x y) more

Yorick

func multiply(x, y) {
    return x * y;
}

Example of interactive usage:

> multiply(2, 4.5)
9

Z80 Assembly

A function's return values are whatever registers or memory are changed by the function. A good programmer will explain what is returned where by using comments.

doMultiply:
;returns HL = HL times A. No overflow protection.
push bc
push de
    rrca                     ;test if A is odd or even by dividing A by 2.
    jr c, isOdd
        ;is even

        ld b,a
loop_multiplyByEvenNumber:
        add hl,hl           ;double A until B runs out.
        djnz loop_multiplyByEvenNumber
pop de
pop bc
ret

isOdd:
    push hl
    pop de                  ;de contains original HL. We'll need it later.
    ld b,a
loop_multiplyByOddNumber:
    add hl,hl
    djnz loop_multiplyByOddNumber
    add hl,de             ;now add in original HL for the leftover add.
pop de
pop bc
ret

zig

fun multiply(x: i64, y: i64) i64 { 
    return x * y;
}

//example call
const x: i64 = 4;
const y: i64 = 23;
_ = multipy(x, y); // --> 93

zkl

fcn multiply(x,y){x*y}
fcn(x,y){x*y}(4.5,3) // --> 13.5

Since all functions are vararg:

fcn multiply{vm.arglist.reduce('*)}
multiply(1,2,3,4,5) //--> 120

Operators are first class objects so:

var mul=Op("*"); mul(4,5) //-->20