How does the random number generator work in C?

问题:

I'm trying to generate random number between 0 and 40(inclusive). So the code I Implemented is this-

 y=rand()%41;

However everytime I click compile and hit Run. It outputs the same random numbers. As in say for instance I ran this in a loop.

for(i=0;i<4;i++)
{
     y=rand()%41;
     printf("%d ",y);
}

Every single time, the output is the same 4 numbers. It always outputs 14,2,etc etc on the terminal. No matter what.

So my first question is, why is this happening?

and Secondly, how does the random number generator work in C?

I thought since I include time.h library, the numbers are generated through some standard algorithm using the system time. And since the system time is continuously changing, the numbers that are generated should also change every time I run the program.

回答1:

rand() generates only pseudorandom numbers. This means that every time you run your code you will get exactly the same sequence of numbers.

Consider using

srand(time(NULL))

to get every time different numbers. In fact a possible implementation for rand is

next = next * 1103515245 + 12345;
return (UINT32)(next>>16) & RAND_MAX;

where next is defined as

static UINT32 next = 1;

Calling srand() has the effect of changing the initial value of next, thus changing the "next" value you get as result.

回答2:

As for how it works, it depends. Many implementations use a Linear Congruential Generator, with different parameters.

回答3:

The algorithm used for rand is unspecified by the C Standard,

By specification, if you don't call srand before a call to rand in your program, it is as if srand(1) was called: the seed value will be 1 at every execution of the program and the generated sequence will be always the same.

A common way to have different seeds for different executions of the program is to use a seed which depends on the current time like:

srand(time(NULL));  

回答4:

This is actually a FAQ on comp.lang.c. Here's the solution that they suggest:

(int)((double)rand() / ((double) RAND_MAX + 1) * N )

Where N is the ceiling of your range of random numbers. This is because the low order bits on bad C compilers are "shockingly non-random". This doesn't get around the need for using srand(). Note, however that srand( time(NULL) ) should be called outside of your loop... time() has a resolution of 1 second, so calling it inside of the loop will re-initialize your random number generator to the same seed many times in a row.

The need for this is probably largely historical, I'm sure that modern compilers probably don't have random number generators which emit really bad random numbers, but I remember writing a program using the Borland C compiler which would cycle through about 5 numbers when I used rand() % 41 repeatedly.

回答5:

Reason is that rand() is using the same seeding everytime you run. You have to seed it yourself. srand ( time(NULL) );` is usually used to initialize random seed.

回答6:

It is generated by an argument passed seed. To generate different numbers add this before calling the rand() function:

srand (time(NULL));

This generates a new random seed.

You should have this library: #include <time.h> And if you still have an error use this one as well: #include <stdlib.h>