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
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
|
function [y]=RndInt(m,n,imin,imax)
// discrete uniform random number
//-------------------------------
// Copyright INRIA
y=rand(m,n,'uniform')
y=int(floor(y*(imax+1-imin)+ imin ));
function [y]=RndDisc(m,n,x,p)
// discrete law random number
// sum p_i delta_{x_i}
//-------------------------------
p1=[0,p];p1=cumsum(p1);
y=rand(m,n,'uniform')
N=prod(size(x));
res=0*ones(m*n);
for i=1:N,z=0*ones(m*n,1),id=find( p1(i) <= y & y < p1(i+1) ),
z(id)=x(i)*ones(prod(size(id))),res=res+z;
end
y=matrix(res,m,n);
function [y]=Binomial(m,n,pb,nb)
// Binomial law (p,N)
// P{X=n} = C_N^n p^n (1-p)^(N-n)
//----------------------------------
res=[];
// we use blocks of size 100 to avoid overflows
ntir=100;
ntirc=ntir;
y=rand(ntir,nb,'uniform');
indy= find( y < pb);
y=0*ones(y);
y(indy)=1;
y=sum(y,'c')
res=[res;y];
while ( ntirc < m*n )
y=rand(ntir,nb,'uniform');
indy= find(y< pb);
y=0*ones(y);
y(indy)=1;
y=sum(y,'c')
res=[res;y];
ntirc=ntirc+ntir;
end
y=matrix(res(1:m*n),m,n);
function [y]=Geom(m,n,p)
// P(0)= 0 P(i) = p*(1-p)^{n-1} P(inf)=0
// E = 1/p ; sig2= (1-p)/p^2
//--------------------------------------
if p >= 1 then write(%io(2),'p must be < 1');end
y=0*ones(m,n)
for i=1:m*n,
samples=1
z=rand(1,1,'uniform');
while( z < 1-p) ,z=rand(1,1,'uniform'); samples=samples+1;end
y(i)= samples;
end
y=matrix(y,m,n)
function [y]=Poisson(m,n,pmean)
// P{n} = exp(-lambda)lambda^n/n!
// pmean =lambda
//----------------------------
y=0*ones(m,n)
bound= exp(-pmean);
for i=1:m*n,
count=0
lprod=1
while( lprod >= bound), lprod=lprod*rand(1,1,'uniform');
count=count+1;end
y(i)=count-1;
end
y=matrix(y,m,n)
function [y]=Exp(m,n,lambda)
// lambda exp(-lambda x) x>=0
// ---------------------------
y=(-1/lambda)* log(rand(m,n,'uniform'));
function [y]=Weibull(m,n,alpha,beta)
//-------------------------------
y=rand(m,n,'uniform')
y= (beta*( - log(1-y)))^(1/alpha)
function [y]=HyperGeom(m,n,Mean,var)
//-------------------------------
z = var / (Mean * Mean);
pP = 0.5 * (1.0 - sqrt((z - 1.0) / ( z + 1.0 )));
y=rand(m,n,'uniform')
zz=find( y > pP) ;
y=pP*ones(y);
y(zz) = (1-pP)*ones(zz);
y1=rand(m,n,'uniform')
y=-Mean * log(y1) ./ (2.0 * y) ;
function [y]=Erlang(m,n,pMean,pVariance)
//-------------------------------
k = int( (pMean * pMean ) / pVariance + 0.5 );
if (k <= 0) then k=1;end
a = k / pMean;
// we use blocks of size 100 to avoid overflows
res=[];
ntir=100;
ntirc=ntir;
y=rand(ntir,k,'uniform');
y= -log(prod(y,'r'))/a;
res=[res;y];
while ( ntirc < m*n )
y=rand(ntir,k,'uniform');
y= -log(prod(y,'r'))/a;
res=[res;y];
ntirc=ntirc+ntir;
end
y=matrix(res(1:m*n),m,n);
function [y]=RndPerm(n)
//-------------------------------
// a uniform random permutation of (1:n)
y=rand(1,n,'uniform')
[us,z]=sort(y);
|
