1 |
Canek Peláez Valdés <caneko@×××××.com> [14-06-06 17:36]: |
2 |
> On Thu, Jun 5, 2014 at 9:56 PM, <meino.cramer@×××.de> wrote: |
3 |
> > Hi, |
4 |
> > |
5 |
> > I am experimenting with the C code of the ISAAC pseudo random number generator |
6 |
> > (http://burtleburtle.net/bob/rand/isaacafa.html). |
7 |
> > |
8 |
> > Currently the implementation creates (on my embedded linux) 32 bit |
9 |
> > hexadecimal output. |
10 |
> |
11 |
> So it's a 32 bit integer. |
12 |
> |
13 |
> > From this I want to create random numbers in the range of [a-Za-z0-9] |
14 |
> > *without violating randomness* and (if possible) without throwing |
15 |
> > away bits of the output. |
16 |
> |
17 |
> You mean *characters* int the range [A-Za-z0-9]? |
18 |
> |
19 |
> > How can I do this mathemtically (in concern of the quality of output) |
20 |
> > correct? |
21 |
> |
22 |
> The easiest thing to do would be: |
23 |
> |
24 |
> ------------------------------------------------------------------------------- |
25 |
> #include <time.h> |
26 |
> #include <stdio.h> |
27 |
> #include <stdlib.h> |
28 |
> |
29 |
> #define N (26+26+10) |
30 |
> |
31 |
> static char S[] = { 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', |
32 |
> 'K', 'L', 'M', |
33 |
> 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', |
34 |
> 'X', 'Y', 'Z', |
35 |
> 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', |
36 |
> 'k', 'l', 'm', |
37 |
> 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', |
38 |
> 'x', 'y', 'z', |
39 |
> '0', '1', '2', '3', '4', '5', '6', '7', '8', '9' }; |
40 |
> |
41 |
> int |
42 |
> next_character() |
43 |
> { |
44 |
> // Use the correct call for ISAAC instead of rand() |
45 |
> unsigned int idx = rand() % N; |
46 |
> return S[idx]; |
47 |
> } |
48 |
> |
49 |
> int |
50 |
> main(int argc, char* argv[]) |
51 |
> { |
52 |
> // Use the correct call for initializing the ISAAC seed |
53 |
> srand((unsigned int)time(NULL)); |
54 |
> for (int i = 0; i < 20; i++) // --std=c99 |
55 |
> printf("%c\n", next_character()); |
56 |
> return 0; |
57 |
> } |
58 |
> ------------------------------------------------------------------------------- |
59 |
> |
60 |
> If the ISAAC RNG has a good distribution, then the next_character() |
61 |
> function will give a good distribution among the set [A-Za-z0-9]. |
62 |
> |
63 |
> Unless I missunderstood what you meant with "create random numbers in |
64 |
> the range of [a-Za-z0-9]". |
65 |
> |
66 |
> Regards. |
67 |
> -- |
68 |
> Canek Peláez Valdés |
69 |
> Profesor de asignatura, Facultad de Ciencias |
70 |
> Universidad Nacional Autónoma de México |
71 |
> |
72 |
|
73 |
Hi, |
74 |
|
75 |
Thank you very much for the input! :) |
76 |
|
77 |
I have a question about the algorithm: |
78 |
Suppose rand() has an equal distribution of numbers and furthermore |
79 |
one has a count of 2^32 random numbers listed in numerical sort |
80 |
order. |
81 |
In this list each number would appear (nearly) with the same count: 1 |
82 |
|
83 |
To get an better imagination of that...suppose the rand() would only |
84 |
return numbers in the range of 1...12 and the alphabet has only 8 |
85 |
characters (as 2^32 is not devideable by 62) |
86 |
|
87 |
rand(): |
88 |
1 2 3 4 5 6 7 8 9 10 11 12 |
89 |
|
90 |
rand()%N : rand()%7 |
91 |
1 2 3 4 5 6 7 0 1 2 3 4 |
92 |
|
93 |
or in other words: An even distribution of numbers of rand() |
94 |
would result in a unevenly distributed sequence of characters...or? |
95 |
This would break the quality of ISAACs output. |
96 |
|
97 |
I am sure I did something wrong here...but where is the logic trap? |
98 |
|
99 |
|
100 |
Thank you very much for any help in advance! |
101 |
Best regards, |
102 |
mcc |