1 |
On Fri, Jun 06, 2014 at 08:39:28PM +0200, meino.cramer@×××.de wrote: |
2 |
> Canek Peláez Valdés <caneko@×××××.com> [14-06-06 17:36]: |
3 |
> > On Thu, Jun 5, 2014 at 9:56 PM, <meino.cramer@×××.de> wrote: |
4 |
> > > Hi, |
5 |
> > > |
6 |
> > > I am experimenting with the C code of the ISAAC pseudo random number generator |
7 |
> > > (http://burtleburtle.net/bob/rand/isaacafa.html). |
8 |
> > > |
9 |
> > > Currently the implementation creates (on my embedded linux) 32 bit |
10 |
> > > hexadecimal output. |
11 |
> > |
12 |
> > So it's a 32 bit integer. |
13 |
> > |
14 |
> > > From this I want to create random numbers in the range of [a-Za-z0-9] |
15 |
> > > *without violating randomness* and (if possible) without throwing |
16 |
> > > away bits of the output. |
17 |
> > |
18 |
> > You mean *characters* int the range [A-Za-z0-9]? |
19 |
> > |
20 |
> > > How can I do this mathemtically (in concern of the quality of output) |
21 |
> > > correct? |
22 |
> > |
23 |
> > The easiest thing to do would be: |
24 |
> > |
25 |
> > ------------------------------------------------------------------------------- |
26 |
> > #include <time.h> |
27 |
> > #include <stdio.h> |
28 |
> > #include <stdlib.h> |
29 |
> > |
30 |
> > #define N (26+26+10) |
31 |
> > |
32 |
> > static char S[] = { 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', |
33 |
> > 'K', 'L', 'M', |
34 |
> > 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', |
35 |
> > 'X', 'Y', 'Z', |
36 |
> > 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', |
37 |
> > 'k', 'l', 'm', |
38 |
> > 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', |
39 |
> > 'x', 'y', 'z', |
40 |
> > '0', '1', '2', '3', '4', '5', '6', '7', '8', '9' }; |
41 |
> > |
42 |
> > int |
43 |
> > next_character() |
44 |
> > { |
45 |
> > // Use the correct call for ISAAC instead of rand() |
46 |
> > unsigned int idx = rand() % N; |
47 |
> > return S[idx]; |
48 |
> > } |
49 |
> > |
50 |
> > int |
51 |
> > main(int argc, char* argv[]) |
52 |
> > { |
53 |
> > // Use the correct call for initializing the ISAAC seed |
54 |
> > srand((unsigned int)time(NULL)); |
55 |
> > for (int i = 0; i < 20; i++) // --std=c99 |
56 |
> > printf("%c\n", next_character()); |
57 |
> > return 0; |
58 |
> > } |
59 |
> > ------------------------------------------------------------------------------- |
60 |
> > |
61 |
> > If the ISAAC RNG has a good distribution, then the next_character() |
62 |
> > function will give a good distribution among the set [A-Za-z0-9]. |
63 |
> > |
64 |
> > Unless I missunderstood what you meant with "create random numbers in |
65 |
> > the range of [a-Za-z0-9]". |
66 |
> > |
67 |
> > Regards. |
68 |
> > -- |
69 |
> > Canek Peláez Valdés |
70 |
> > Profesor de asignatura, Facultad de Ciencias |
71 |
> > Universidad Nacional Autónoma de México |
72 |
> > |
73 |
> |
74 |
> Hi, |
75 |
> |
76 |
> Thank you very much for the input! :) |
77 |
> |
78 |
> I have a question about the algorithm: |
79 |
> Suppose rand() has an equal distribution of numbers and furthermore |
80 |
> one has a count of 2^32 random numbers listed in numerical sort |
81 |
> order. |
82 |
> In this list each number would appear (nearly) with the same count: 1 |
83 |
> |
84 |
> To get an better imagination of that...suppose the rand() would only |
85 |
> return numbers in the range of 1...12 and the alphabet has only 8 |
86 |
> characters (as 2^32 is not devideable by 62) |
87 |
> |
88 |
> rand(): |
89 |
> 1 2 3 4 5 6 7 8 9 10 11 12 |
90 |
> |
91 |
> rand()%N : rand()%7 |
92 |
> 1 2 3 4 5 6 7 0 1 2 3 4 |
93 |
> |
94 |
> or in other words: An even distribution of numbers of rand() |
95 |
> would result in a unevenly distributed sequence of characters...or? |
96 |
> This would break the quality of ISAACs output. |
97 |
> |
98 |
> I am sure I did something wrong here...but where is the logic trap? |
99 |
> |
100 |
|
101 |
This is the thing I explained in my message. |
102 |
|
103 |
-- |
104 |
-Matti |