Random Function

[zodiac976]

thats what you call spin. you should be in politics.

People like to see stuff done in different ways
if everyone programmed the same that would
be very bad and I hate politics. I will stick with
programming .

[MarauderIIC]

Top Secret level encryption,

I would be willing to bet that this doesn't use rand().

[zodiac976]

Top Secret level encryption,

I would be willing to bet that this doesn't use rand().

Something more sophisticated in my opinion.

[avansc]

Top Secret level encryption,

I would be willing to bet that this doesn't use rand().

Something more sophisticated in my opinion.

like?

[zodiac976]

In college I had to learn about randomizing on my own
and put it in my mid-term and scored good on it.

I wanted to post my own version of how I did it even
though it isn't the absolute perfect solution .

[zodiac976]

Top Secret level encryption,

I would be willing to bet that this doesn't use rand().

Something more sophisticated in my opinion.

like?

Don't ask me I am not that smart .

[MarauderIIC]

like?

Like some calculation of Pi.

Oh, and to sum it up for the OP:

Is there a way to have a random range function that never repeats any digit? :|

No. If you don't reseed by calling srand(), they seem to repeat after a few calls to rand(). If you do reseed, and you use time(NULL) to seed with every time, you'll notice more repetition than seeding with the previously generated random number, but it will eventually repeat, either way.

Look at it this way, unsigned ints have a range of 0-4,294,967,295
If you call rand() 4,294,967,297 times, you will have at least one number repeat. But since I doubt that you want 4,294,967,296 unique random numbers -- rather, you probably only want like, 10 -- then after 10 calls, you are guaranteed to repeat. (This is the common-sense Pigeonhole principle, which you can do all sorts of fun things with. For example, if the maximum hairs on a head is 1 million and there are 1 .5 million people in London, then at least 2 people in London have the exact same number of hairs on their head).

In a perfect world, as you approach the 5th number, you'll have a 50% chance of your 6th number being a repeat. As you approach the 9th number, you have a 90% chance of your 10th number being a repeat. But since computers are psuedo-random, as opposed to truly random, your chances of getting a duplicate result always seem to be higher. You can reduce the chance of getting a duplicate by seeding with time(NULL) on the first call and reseeding with the last result of rand() [pure, not modded down to your range] on each subsequent call to rand().

Hopefully that totally answers the question.

[zodiac976]

like?

Like some calculation of Pi.

Oh, and to sum it up for the OP:

Is there a way to have a random range function that never repeats any digit?

No. If you don't reseed by calling srand(), they seem to repeat after a few calls to rand(). If you do reseed, and you use time(NULL) to seed with every time, you'll notice more repetition than seeding with the previously generated random number, but it will eventually repeat, either way.

Look at it this way, unsigned ints have a range of 0-4,294,967,295
If you call rand() 4,294,967,297 times, you will have at least one number repeat. But since I doubt that you want 4,294,967,296 unique random numbers -- rather, you probably only want like, 10 -- then after 10 calls, you are guaranteed to repeat. (This is the common-sense Pigeonhole principle, which you can do all sorts of fun things with. For example, if the maximum hairs on a head is 1 million and there are 1 .5 million people in London, then at least 2 people in London have the exact same number of hairs on their head).

In a perfect world, as you approach the 5th number, you'll have a 50% chance of your 6th number being a repeat. As you approach the 9th number, you have a 90% chance of your 10th number being a repeat. But since computers are psuedo-random, as opposed to truly random, your chances of getting a duplicate result always seem to be higher. You can reduce the chance of getting a duplicate by seeding with time(NULL) on the first call and reseeding with the last result of rand() [pure, not modded down to your range] on each subsequent call to rand().

Hopefully that totally answers the question.

Well said and I didn't know about reseeding again .

[MarauderIIC]

Well said and I didn't know about reseeding again ;).

Thanks and http://elysianshadows.com/phpBB3/viewto ... 303#p42303
Which is why avansc asked if you read, I think, since your answer was nearly spot-on to what I had already written.

[zodiac976]

Well said and I didn't know about reseeding again .

Thanks and http://elysianshadows.com/phpBB3/viewto ... 303#p42303
Which is why avansc asked if you read, I think, since your answer was nearly spot-on to what I had already written.

Sorry, I will avoid it next time.

[avansc]

funny enough the biggest thing in cryptography is not random numbers, but primes.
come to think of it. random plays little to no role in cryptography.

like DES, MD-n, SHA.. all those dont have any randomization in them.

[dandymcgee]

Top Secret level encryption,

I would be willing to bet that this doesn't use rand().

Depending on the organization.. I'll take that bet.

[MarauderIIC]

You said Top Secret.

[Ginto8]

another (you could say more random) way of seeding it is something like this:

Code: Select all

srand((unsigned)time(NULL));
srand(rand()%3000);
srand(rand()%3000);

if you do that before each random number, it would be a (little bit of a) waste of CPU power, but it would probably make it seem a little more random.

[Scoody]

another (you could say more random) way of seeding it is something like this:

Code: Select all

srand((unsigned)time(NULL));
srand(rand()%3000);
srand(rand()%3000);

if you do that before each random number, it would be a (little bit of a) waste of CPU power, but it would probably make it seem a little more random.

It's just a waste of CPU time.
rand() creates the same "random" numbers if seeded the same number; so seeding rand() with a rand() would yield the same numbers anyway.
My point being, it won't be any more random :p