I use standard C random number generation logic in my application.
-------------------------------------------------
long nCounter;
long lRndNo;
char rand[11];

srand48(nCounter);
lRndNo = lrand48();
sprintf(rand,"%010.10d",lRndNo);

--------------------------------------------------

However we always find that the rand generated starts with '0','1' or '2'. I could not logically prove if is guaranteed to start with these. We are designing a new program with the assumption that it will not start with '5'.

Can someone please explain how this logic (linear conguential) works to start with '0','1','2'?
I will highly appreciate.
Thanks
Asutosh

Please don't post the same question twice. I deleted the duplicate.

The largest number in a 32 bit signed integer is 2,147,483,647. So if a 32 bit signed integer has more than 9 significant digits, it will have exactly 10 and the most significant digit will be a 1 or a 2. There is no way to store a larger number in 32 bits.

So you are building in an assumption of 32 bit arithmetic. That sounds dangerous today. There are quite a few 64 bit machines around.

Or the other hand, you are calling srand48(). What makes srand48() cool is that it uses 48 bit aritmetic internally and returns a 32 bit quantity. An assumpion of 32 bit arithmetic is built heavily into srand48(), even up to its name. So maybe you can get away with it.