xref: /openbsd/lib/libc/stdlib/rand48.3 (revision db3296cf)
1.\" Copyright (c) 1993 Martin Birgmeier
2.\" All rights reserved.
3.\"
4.\" You may redistribute unmodified or modified versions of this source
5.\" code provided that the above copyright notice and this and the
6.\" following conditions are retained.
7.\"
8.\" This software is provided ``as is'', and comes with no warranties
9.\" of any kind. I shall in no event be liable for anything that happens
10.\" to anyone/anything when using this software.
11.\"
12.\"	$OpenBSD: rand48.3,v 1.10 2003/06/01 19:27:27 jmc Exp $
13.\"
14.Dd October 8, 1993
15.Dt RAND48 3
16.Os
17.Sh NAME
18.Nm drand48 ,
19.Nm erand48 ,
20.Nm lrand48 ,
21.Nm nrand48 ,
22.Nm mrand48 ,
23.Nm jrand48 ,
24.Nm srand48 ,
25.Nm seed48 ,
26.Nm lcong48
27.Nd pseudo-random number generators and initialization routines
28.Sh SYNOPSIS
29.Fd #include <stdlib.h>
30.Ft double
31.Fn drand48 void
32.Ft double
33.Fn erand48 "unsigned short xseed[3]"
34.Ft long
35.Fn lrand48 void
36.Ft long
37.Fn nrand48 "unsigned short xseed[3]"
38.Ft long
39.Fn mrand48 void
40.Ft long
41.Fn jrand48 "unsigned short xseed[3]"
42.Ft void
43.Fn srand48 "long seed"
44.Ft "unsigned short *"
45.Fn seed48 "unsigned short xseed[3]"
46.Ft void
47.Fn lcong48 "unsigned short p[7]"
48.Sh DESCRIPTION
49The
50.Fn rand48
51family of functions generates pseudo-random numbers using a linear
52congruential algorithm working on integers 48 bits in size.
53The particular formula employed is
54r(n+1) = (a * r(n) + c) mod m
55where the default values are
56for the multiplicand a = 0xfdeece66d = 25214903917 and
57the addend c = 0xb = 11.
58The modulus is always fixed at m = 2 ** 48.
59r(n) is called the seed of the random number generator.
60.Pp
61For all the six generator routines described next, the first
62computational step is to perform a single iteration of the algorithm.
63.Pp
64.Fn drand48
65and
66.Fn erand48
67return values of type double.
68The full 48 bits of r(n+1) are
69loaded into the mantissa of the returned value, with the exponent set
70such that the values produced lie in the interval [0.0, 1.0].
71.Pp
72.Fn lrand48
73and
74.Fn nrand48
75return values of type long in the range
76[0, 2**31-1].
77The high-order (31) bits of
78r(n+1) are loaded into the lower bits of the returned value, with
79the topmost (sign) bit set to zero.
80.Pp
81.Fn mrand48
82and
83.Fn jrand48
84return values of type long in the range
85[-2**31, 2**31-1].
86The high-order (32) bits of r(n+1) are loaded into the returned value.
87.Pp
88.Fn drand48 ,
89.Fn lrand48 ,
90and
91.Fn mrand48
92use an internal buffer to store r(n).
93For these functions
94the initial value of r(0) = 0x1234abcd330e = 20017429951246.
95.Pp
96On the other hand,
97.Fn erand48 ,
98.Fn nrand48 ,
99and
100.Fn jrand48
101use a user-supplied buffer to store the seed r(n),
102which consists of an array of 3 shorts, where the zeroth member
103holds the least significant bits.
104.Pp
105All functions share the same multiplicand and addend.
106.Pp
107.Fn srand48
108is used to initialize the internal buffer r(n) of
109.Fn drand48 ,
110.Fn lrand48 ,
111and
112.Fn mrand48
113such that the 32 bits of the seed value are copied into the upper 32 bits
114of r(n), with the lower 16 bits of r(n) arbitrarily being set to 0x330e.
115Additionally, the constant multiplicand and addend of the algorithm are
116reset to the default values given above.
117.Pp
118.Fn seed48
119also initializes the internal buffer r(n) of
120.Fn drand48 ,
121.Fn lrand48 ,
122and
123.Fn mrand48 ,
124but here all 48 bits of the seed can be specified in an array of 3 shorts,
125where the zeroth member specifies the lowest bits.
126Again, the constant multiplicand and addend of the algorithm are
127reset to the default values given above.
128.Fn seed48
129returns a pointer to an array of 3 shorts which contains the old seed.
130This array is statically allocated, so its contents are lost after
131each new call to
132.Fn seed48 .
133.Pp
134Finally,
135.Fn lcong48
136allows full control over the multiplicand and addend used in
137.Fn drand48 ,
138.Fn erand48 ,
139.Fn lrand48 ,
140.Fn nrand48 ,
141.Fn mrand48 ,
142and
143.Fn jrand48 ,
144and the seed used in
145.Fn drand48 ,
146.Fn lrand48 ,
147and
148.Fn mrand48 .
149An array of 7 shorts is passed as parameter; the first three shorts are
150used to initialize the seed; the second three are used to initialize the
151multiplicand; and the last short is used to initialize the addend.
152It is thus not possible to use values greater than 0xffff as the addend.
153.Pp
154Note that all three methods of seeding the random number generator
155always also set the multiplicand and addend for any of the six
156generator calls.
157.Pp
158For a more powerful random number generator, see
159.Xr random 3 .
160.Sh SEE ALSO
161.Xr arc4random 3 ,
162.Xr rand 3 ,
163.Xr random 3
164.Sh AUTHORS
165Martin Birgmeier
166