blob: b2c6f76d3dbfb1cc65670621aad721971f34617a (
plain) (
blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
|
REAL FUNCTION sla_GRESID (S)
*+
* - - - - - - -
* G R E S I D
* - - - - - - -
*
* Generate pseudo-random normal deviate ( = 'Gaussian residual')
* (single precision)
*
* Given:
* S real standard deviation
*
* The results of many calls to this routine will be
* normally distributed with mean zero and standard deviation S.
*
* The Box-Muller algorithm is used. This is described in
* Numerical Recipes, section 7.2.
*
* !!! Microsoft Fortran dependent - calls the RAN routine !!!
* !!! To seed the random-number generator, either call the !!!
* !!! Microsoft SEED routine specifying some INTEGER*2 !!!
* !!! seed or call the function sla_RANDOM specifying some !!!
* !!! REAL seed. !!!
*
* P.T.Wallace Starlink 1 April 1993
*
* Copyright (C) 1995 Rutherford Appleton Laboratory
*-
IMPLICIT NONE
REAL S
REAL X,Y,RV,R,W,GNEXT,G
LOGICAL FIRST
SAVE GNEXT,RV,FIRST
DATA FIRST / .TRUE. /
* Second normal deviate of the pair available?
IF (FIRST) THEN
* No - generate two random numbers in range +/- 1
1 CONTINUE
CALL RANDOM(RV) !!! PC
X = 2.0*RV-1.0
CALL RANDOM(RV) !!! PC
Y = 2.0*RV-1.0
* Try again if not in unit circle
R = X*X+Y*Y
IF (R.GE.1.0) GO TO 1
* Box-Muller transformation, generating two deviates
W = SQRT(-2.0*LOG(R)/MAX(R,1E-20))
GNEXT = X*W
G = Y*W
* Set flag to indicate availability of next deviate
FIRST = .FALSE.
ELSE
* Return second deviate of the pair & reset flag
G = GNEXT
FIRST = .TRUE.
END IF
* Scale the deviate by the required standard deviation
sla_GRESID = G*S
END
|
