rand48.3 revision 1.1.2.1 1 1.1 jtc \" Copyright (c) 1993 Martin Birgmeier
2 1.1 jtc .\" All rights reserved.
3 1.1 jtc .\"
4 1.1 jtc .\" You may redistribute unmodified or modified versions of this source
5 1.1 jtc .\" code provided that the above copyright notice and this and the
6 1.1 jtc .\" following conditions are retained.
7 1.1 jtc .\"
8 1.1 jtc .\" This software is provided ``as is'', and comes with no warranties
9 1.1 jtc .\" of any kind. I shall in no event be liable for anything that happens
10 1.1 jtc .\" to anyone/anything when using this software.
11 1.1 jtc .\"
12 1.1.2.1 mycroft .\" $Id: rand48.3,v 1.1.2.1 1994/10/06 04:36:12 mycroft Exp $
13 1.1 jtc .\"
14 1.1 jtc .Dd October 8, 1993
15 1.1 jtc .Dt RAND48 3
16 1.1 jtc .Os
17 1.1 jtc .Sh NAME
18 1.1 jtc .Nm drand48 ,
19 1.1 jtc .Nm erand48 ,
20 1.1 jtc .Nm lrand48 ,
21 1.1 jtc .Nm nrand48 ,
22 1.1 jtc .Nm mrand48 ,
23 1.1 jtc .Nm jrand48 ,
24 1.1 jtc .Nm srand48 ,
25 1.1 jtc .Nm seed48 ,
26 1.1 jtc .Nm lcong48
27 1.1 jtc .Nd pseudo random number generators and initialization routines
28 1.1 jtc .Sh SYNOPSIS
29 1.1 jtc .Fd #include <stdlib.h>
30 1.1 jtc .Ft double
31 1.1 jtc .Fn drand48 void
32 1.1 jtc .Ft double
33 1.1 jtc .Fn erand48 "unsigned short xseed[3]"
34 1.1 jtc .Ft long
35 1.1 jtc .Fn lrand48 void
36 1.1 jtc .Ft long
37 1.1 jtc .Fn nrand48 "unsigned short xseed[3]"
38 1.1 jtc .Ft long
39 1.1 jtc .Fn mrand48 void
40 1.1 jtc .Ft long
41 1.1 jtc .Fn jrand48 "unsigned short xseed[3]"
42 1.1 jtc .Ft void
43 1.1 jtc .Fn srand48 "long seed"
44 1.1 jtc .Ft "unsigned short *"
45 1.1 jtc .Fn seed48 "unsigned short xseed[3]"
46 1.1 jtc .Ft void
47 1.1 jtc .Fn lcong48 "unsigned short p[7]"
48 1.1 jtc .Sh DESCRIPTION
49 1.1 jtc The
50 1.1 jtc .Fn rand48
51 1.1 jtc family of functions generates pseudo-random numbers using a linear
52 1.1 jtc congruential algorithm working on integers 48 bits in size. The
53 1.1 jtc particular formula employed is
54 1.1 jtc r(n+1) = (a * r(n) + c) mod m
55 1.1 jtc where the default values are
56 1.1 jtc for the multiplicand a = 0xfdeece66d = 25214903917 and
57 1.1 jtc the addend c = 0xb = 11. The modulus is always fixed at m = 2 ** 48.
58 1.1 jtc r(n) is called the seed of the random number generator.
59 1.1 jtc .Pp
60 1.1 jtc For all the six generator routines described next, the first
61 1.1 jtc computational step is to perform a single iteration of the algorithm.
62 1.1 jtc .Pp
63 1.1 jtc .Fn drand48
64 1.1 jtc and
65 1.1 jtc .Fn erand48
66 1.1 jtc return values of type double. The full 48 bits of r(n+1) are
67 1.1 jtc loaded into the mantissa of the returned value, with the exponent set
68 1.1 jtc such that the values produced lie in the interval [0.0, 1.0).
69 1.1 jtc .Pp
70 1.1 jtc .Fn lrand48
71 1.1 jtc and
72 1.1 jtc .Fn nrand48
73 1.1 jtc return values of type long in the range
74 1.1.2.1 mycroft [0, 2**31-1]. The high-order (31) bits of
75 1.1.2.1 mycroft r(n+1) are loaded into the lower bits of the returned value, with
76 1.1.2.1 mycroft the topmost (sign) bit set to zero.
77 1.1 jtc .Pp
78 1.1 jtc .Fn mrand48
79 1.1 jtc and
80 1.1 jtc .Fn jrand48
81 1.1 jtc return values of type long in the range
82 1.1.2.1 mycroft [-2**31, 2**31-1]. The high-order (32) bits of
83 1.1.2.1 mycroft r(n+1) are loaded into the returned value.
84 1.1 jtc .Pp
85 1.1 jtc .Fn drand48 ,
86 1.1 jtc .Fn lrand48 ,
87 1.1 jtc and
88 1.1 jtc .Fn mrand48
89 1.1 jtc use an internal buffer to store r(n). For these functions
90 1.1 jtc the initial value of r(0) = 0x1234abcd330e = 20017429951246.
91 1.1 jtc .Pp
92 1.1 jtc On the other hand,
93 1.1 jtc .Fn erand48 ,
94 1.1 jtc .Fn nrand48 ,
95 1.1 jtc and
96 1.1 jtc .Fn jrand48
97 1.1 jtc use a user-supplied buffer to store the seed r(n),
98 1.1 jtc which consists of an array of 3 shorts, where the zeroth member
99 1.1 jtc holds the least significant bits.
100 1.1 jtc .Pp
101 1.1 jtc All functions share the same multiplicand and addend.
102 1.1 jtc .Pp
103 1.1 jtc .Fn srand48
104 1.1 jtc is used to initialize the internal buffer r(n) of
105 1.1 jtc .Fn drand48 ,
106 1.1 jtc .Fn lrand48 ,
107 1.1 jtc and
108 1.1 jtc .Fn mrand48
109 1.1 jtc such that the 32 bits of the seed value are copied into the upper 32 bits
110 1.1 jtc of r(n), with the lower 16 bits of r(n) arbitrarily being set to 0x330e.
111 1.1 jtc Additionally, the constant multiplicand and addend of the algorithm are
112 1.1 jtc reset to the default values given above.
113 1.1 jtc .Pp
114 1.1 jtc .Fn seed48
115 1.1 jtc also initializes the internal buffer r(n) of
116 1.1 jtc .Fn drand48 ,
117 1.1 jtc .Fn lrand48 ,
118 1.1 jtc and
119 1.1 jtc .Fn mrand48 ,
120 1.1 jtc but here all 48 bits of the seed can be specified in an array of 3 shorts,
121 1.1 jtc where the zeroth member specifies the lowest bits. Again,
122 1.1 jtc the constant multiplicand and addend of the algorithm are
123 1.1 jtc reset to the default values given above.
124 1.1 jtc .Fn seed48
125 1.1 jtc returns a pointer to an array of 3 shorts which contains the old seed.
126 1.1 jtc This array is statically allocated, thus its contents are lost after
127 1.1 jtc each new call to
128 1.1 jtc .Fn seed48 .
129 1.1 jtc .Pp
130 1.1 jtc Finally,
131 1.1 jtc .Fn lcong48
132 1.1 jtc allows full control over the multiplicand and addend used in
133 1.1 jtc .Fn drand48 ,
134 1.1 jtc .Fn erand48 ,
135 1.1 jtc .Fn lrand48 ,
136 1.1 jtc .Fn nrand48 ,
137 1.1 jtc .Fn mrand48 ,
138 1.1 jtc and
139 1.1 jtc .Fn jrand48 ,
140 1.1 jtc and the seed used in
141 1.1 jtc .Fn drand48 ,
142 1.1 jtc .Fn lrand48 ,
143 1.1 jtc and
144 1.1 jtc .Fn mrand48 .
145 1.1 jtc An array of 7 shorts is passed as parameter; the first three shorts are
146 1.1 jtc used to initialize the seed; the second three are used to initialize the
147 1.1 jtc multiplicand; and the last short is used to initialize the addend.
148 1.1 jtc It is thus not possible to use values greater than 0xffff as the addend.
149 1.1 jtc .Pp
150 1.1 jtc Note that all three methods of seeding the random number generator
151 1.1 jtc always also set the multiplicand and addend for any of the six
152 1.1 jtc generator calls.
153 1.1 jtc .Pp
154 1.1 jtc For a more powerful random number generator, see
155 1.1 jtc .Xr random 3
156 1.1 jtc .Sh AUTHOR
157 1.1 jtc Martin Birgmeier
158 1.1 jtc .Sh SEE ALSO
159 1.1 jtc .Xr rand 3 ,
160 1.1 jtc .Xr random 3 .
161