The Random interface

A Random.T (or just a generator) is a pseudo-random number generator. pseudo-random number

INTERFACE Random;

TYPE
  T = OBJECT METHODS
    integer(min := FIRST(INTEGER);
      max := LAST(INTEGER)): INTEGER;
    real(min := 0.0e+0; max := 1.0e+0): REAL;
    longreal(min := 0.0d+0; max := 1.0d+0): LONGREAL;
    extended(min := 0.0x+0; max := 1.0x+0): EXTENDED;
    boolean(): BOOLEAN
  END;
  Default <: T OBJECT METHODS
    init(fixed := FALSE): Default
  END;
END Random.

Individual generators are unmonitored, and all the operations have side effects.

The methods provided by a generator rand are:

The call rand.integer(a, b) returns a uniformly distributed INTEGER in the closed interval [a..b].

The call rand.real(a, b) returns a uniformly distributed REAL in the half-open interval [a..b).

The call longreal and extended are like real, but return values of the specified types.

The call rand.boolean() returns a random BOOLEAN value.

It is a checked runtime error if min > max on any call.

NEW(Default).init() creates and initializes a generator (see below for implementation details). If fixed is TRUE, a predetermined sequence is used. If fixed is FALSE, init chooses a random seed in such a way that different sequences result even if init is called many times in close proximity.

Example.

A good pseudo-random permutation of an array a can be generated as follows:

 WITH rand = NEW(Random.Default).init() DO
   FOR i := FIRST(a) TO LAST(a) - 1 DO
     WITH j = rand.integer(i, LAST(a)) DO
       Exchange "a[i]" and "a[j]"
     END
   END
 END

SRC Modula-3 implementation details.

The object returned by a call of New(Default).init uses an additive generator based on Knuth's Algorithm 3.2.2A (see [Knuth:Vol2] ).