I didnt notice this about random :(

I do something completely different-
I include <cstdlib>, which has a rand() function, and include<ctime>.
I then use the clock() function to execute rand() many times based on how many clock ticks have passed since the program started, which puts me into an entirely random place in the rand()'s random number list.

I had no idea about srand() and all that.

Anyway; cstdlib's rand() doesn't go up in value every time, and a way of making a reliable min and max is by dividing the random number by RAND_MAX, multiplying by (MAX-MIN) and adding MIN.

rand() is not sufficient for most programs. There are several algos available on the internet and in books that are far better and quite simple to implement.

Quote:

---

Originally Posted by Bubba

rand() is not sufficient for most programs. There are several algos available on the internet and in books that are far better and quite simple to implement.

---

The C standard mandates no specific algorithm for rand(). I don't think it's justified to claim that rand() is "not sufficient" when it isn't even fully specified how it works.

I wouldn't use rand() for anything that had statistical significance, since I have no idea how it works (nor should I rely on such knowledge since it can change at any time). That doesn't mean it sucks by default, though.

Anyway, C++ TR1 and C++0x define an extensive library of random number generators that are suitable for serious statistical use. Boost.Random is the library the new standard library is based on, but there are some changes. So if your compiler doesn't provide a TR1 implementation, you can use Boost.Random.

Quote:

---

Originally Posted by Eldarion

I do something completely different-
I include <cstdlib>, which has a rand() function, and include<ctime>.
I then use the clock() function to execute rand() many times based on how many clock ticks have passed since the program started, which puts me into an entirely random place in the rand()'s random number list.

I had no idea about srand() and all that.

Anyway; cstdlib's rand() doesn't go up in value every time, and a way of making a reliable min and max is by dividing the random number by RAND_MAX, multiplying by (MAX-MIN) and adding MIN.

---

Are you sure there is any guarantee that clock() will return values that are random enough.

Anyway, if you keep giving ascending values to srand, there's no surprise that you might get ascending values from the next call to rand(). The whole point of rand() is that there is a long-long randomish mathematical sequence and srand() determines your starting point in the sequence.

This doesn't mean that the first number you get after seeding is particularly unpredictable, e.g

Code:

---

#include <cstdio>
#include <cstdlib>

int main()
{
    for (int seed = 32; seed <= 42; ++seed) {
        srand(seed);
        printf("%d %d\n", seed, rand());
    }
}

---

Quote:

---

32 143
33 146
34 149
35 152
36 156
37 159
38 162
39 165
40 169
41 172
42 175

---

Also see how the (pseudo)randomness kicks in if you draw more random numbers from a seed:

Code:

---

#include <cstdio>
#include <cstdlib>

int main()
{
    for (int seed = 32; seed <= 42; ++seed) {
        srand(seed);
        printf("%d", seed);
        for (int i = 0; i != 8; ++i) {
            printf(" %6d", rand());
        }
        puts("");
    }
}

---

Code:

---

32    143  23988  3067  18799  11241  6380  9660  4148
33    146  1968  20932  10094  21556  10307  12772  1221
34    149  12717  6028  1389  31870  14234  15884  31062
35    152  23465  23892  25453  9417  18160  18997  28135
36    156  1446  8988  16748  19731  22087  22109  25207
37    159  12194  26852  8043  30046  26014  25222  22280
38    162  22942  11948  32107  7593  29941  28334  19353
39    165    923  29813  23402  17907  1099  31447  16426
40    169  11671  14909  14697  28222  5026  1791  13498
41    172  22420      5  5992  5768  8953  4904  10571
42    175    400  17869  30056  16083  12879  8016  7644

---