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
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
|
/*
* Experimental data distribution table generator
* Taken from the uncopyrighted NISTnet code (public domain).
*
* Read in a series of "random" data values, either
* experimentally or generated from some probability distribution.
* From this, create the inverse distribution table used to approximate
* the distribution.
*/
#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#include <malloc.h>
#include <string.h>
#include <sys/types.h>
#include <sys/stat.h>
double *
readdoubles(FILE *fp, int *number)
{
struct stat info;
double *x;
int limit;
int n=0, i;
if (!fstat(fileno(fp), &info) &&
info.st_size > 0) {
limit = 2*info.st_size/sizeof(double); /* @@ approximate */
} else {
limit = 10000;
}
x = calloc(limit, sizeof(double));
if (!x) {
perror("double alloc");
exit(3);
}
for (i=0; i<limit; ++i){
if (fscanf(fp, "%lf", &x[i]) != 1 ||
feof(fp))
break;
++n;
}
*number = n;
return x;
}
void
arraystats(double *x, int limit, double *mu, double *sigma, double *rho)
{
int n=0, i;
double sumsquare=0.0, sum=0.0, top=0.0;
double sigma2=0.0;
for (i=0; i<limit; ++i){
sumsquare += x[i]*x[i];
sum += x[i];
++n;
}
*mu = sum/(double)n;
*sigma = sqrt((sumsquare - (double)n*(*mu)*(*mu))/(double)(n-1));
for (i=1; i < n; ++i){
top += ((double)x[i]- *mu)*((double)x[i-1]- *mu);
sigma2 += ((double)x[i-1] - *mu)*((double)x[i-1] - *mu);
}
*rho = top/sigma2;
}
/* Create a (normalized) distribution table from a set of observed
* values. The table is fixed to run from (as it happens) -4 to +4,
* with granularity .00002.
*/
#define TABLESIZE 16384/4
#define TABLEFACTOR 8192
#ifndef MINSHORT
#define MINSHORT -32768
#define MAXSHORT 32767
#endif
/* Since entries in the inverse are scaled by TABLEFACTOR, and can't be bigger
* than MAXSHORT, we don't bother looking at a larger domain than this:
*/
#define DISTTABLEDOMAIN ((MAXSHORT/TABLEFACTOR)+1)
#define DISTTABLEGRANULARITY 50000
#define DISTTABLESIZE (DISTTABLEDOMAIN*DISTTABLEGRANULARITY*2)
static int *
makedist(double *x, int limit, double mu, double sigma)
{
int *table;
int i, index, first=DISTTABLESIZE, last=0;
double input;
table = calloc(DISTTABLESIZE, sizeof(int));
if (!table) {
perror("table alloc");
exit(3);
}
for (i=0; i < limit; ++i) {
/* Normalize value */
input = (x[i]-mu)/sigma;
index = (int)rint((input+DISTTABLEDOMAIN)*DISTTABLEGRANULARITY);
if (index < 0) index = 0;
if (index >= DISTTABLESIZE) index = DISTTABLESIZE-1;
++table[index];
if (index > last)
last = index +1;
if (index < first)
first = index;
}
return table;
}
/* replace an array by its cumulative distribution */
static void
cumulativedist(int *table, int limit, int *total)
{
int accum=0;
while (--limit >= 0) {
accum += *table;
*table++ = accum;
}
*total = accum;
}
static short *
inverttable(int *table, int inversesize, int tablesize, int cumulative)
{
int i, inverseindex, inversevalue;
short *inverse;
double findex, fvalue;
inverse = (short *)malloc(inversesize*sizeof(short));
for (i=0; i < inversesize; ++i) {
inverse[i] = MINSHORT;
}
for (i=0; i < tablesize; ++i) {
findex = ((double)i/(double)DISTTABLEGRANULARITY) - DISTTABLEDOMAIN;
fvalue = (double)table[i]/(double)cumulative;
inverseindex = (int)rint(fvalue*inversesize);
inversevalue = (int)rint(findex*TABLEFACTOR);
if (inversevalue <= MINSHORT) inversevalue = MINSHORT+1;
if (inversevalue > MAXSHORT) inversevalue = MAXSHORT;
if (inverseindex >= inversesize) inverseindex = inversesize- 1;
inverse[inverseindex] = inversevalue;
}
return inverse;
}
/* Run simple linear interpolation over the table to fill in missing entries */
static void
interpolatetable(short *table, int limit)
{
int i, j, last, lasti = -1;
last = MINSHORT;
for (i=0; i < limit; ++i) {
if (table[i] == MINSHORT) {
for (j=i; j < limit; ++j)
if (table[j] != MINSHORT)
break;
if (j < limit) {
table[i] = last + (i-lasti)*(table[j]-last)/(j-lasti);
} else {
table[i] = last + (i-lasti)*(MAXSHORT-last)/(limit-lasti);
}
} else {
last = table[i];
lasti = i;
}
}
}
static void
printtable(const short *table, int limit)
{
int i;
printf("# This is the distribution table for the experimental distribution.\n");
for (i=0 ; i < limit; ++i) {
printf("%d%c", table[i],
(i % 8) == 7 ? '\n' : ' ');
}
}
int
main(int argc, char **argv)
{
FILE *fp;
double *x;
double mu, sigma, rho;
int limit;
int *table;
short *inverse;
int total;
if (argc > 1) {
if (!(fp = fopen(argv[1], "r"))) {
perror(argv[1]);
exit(1);
}
} else {
fp = stdin;
}
x = readdoubles(fp, &limit);
if (limit <= 0) {
fprintf(stderr, "Nothing much read!\n");
exit(2);
}
arraystats(x, limit, &mu, &sigma, &rho);
#ifdef DEBUG
fprintf(stderr, "%d values, mu %10.4f, sigma %10.4f, rho %10.4f\n",
limit, mu, sigma, rho);
#endif
table = makedist(x, limit, mu, sigma);
free((void *) x);
cumulativedist(table, DISTTABLESIZE, &total);
inverse = inverttable(table, TABLESIZE, DISTTABLESIZE, total);
interpolatetable(inverse, TABLESIZE);
printtable(inverse, TABLESIZE);
return 0;
}
|
