My random number generator. (Looking for feedback)

**Yonut** · 04-20-2019

I have come up with a random number generator. Can someone tell me if this is good or bad and why?

I basically take the nanoseconds from system time and modulus it with the users max value. Adjusting for starting point.

Code:

#include <stdio.h>
#include <libgen.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>
#include <time.h>

// Declare defines.
#define PROGRAM_NAME_SIZE   50

// Declare function prototypes.
long random_number (long max, int start);

// Declare global variables.
char program_name[PROGRAM_NAME_SIZE + 1] = {0};

int main (int argc, char *argv[])
{
// Get program name for error reporting.
    strcpy(program_name, basename(argv[0]));

// Check for correct number of arguments.
    if(argc != 3)
    {
        fprintf(stderr, "usage: %s MAX_VALUE START\n", program_name);
        fprintf(stderr, "%s: START is either 0 or 1\n", program_name);
        exit(EXIT_FAILURE);
    }

// Get random numbers forever.
    while(1)
        printf("%ld\n", random_number(atol(argv[1]), atoi(argv[2])));

// Exit gracefully.
    exit(EXIT_SUCCESS);
}

long random_number (long max, int start)
{
// Declare variables.
    long number = {0};
    struct timespec tp = {0};

// Get the current time.
    if(clock_gettime(CLOCK_REALTIME, &tp) == 1)
    {
        fprintf(stderr, "%s: random_number error: failed (%s)\n", program_name, strerror(errno));
        exit(EXIT_FAILURE);
    }

// Calculate the random number.
    number = tp.tv_nsec % max;

// Change number depending on start.
    if(start == 1)
        if(number == 0)
            number = max;

// Return the number.
    return(number);
}

**Salem** · 04-20-2019

> Can someone tell me if this is good or bad and why?
It's awful, truly awful.

Seriously, don't even suggest this is a random number generator.

It isn't.

Code:

    int histogram[100] = { 0 };
    for ( int i = 0 ; i < 1000000 ; i++ ) {
        long n = random_number(100,0);
        histogram[n]++;
    }
    for ( int i = 0 ; i < 100 ; i++ ) {
        printf("%d %d\n", i, histogram[i]);
    }

Results in

Code:

Picking a modulus which is congruent to the cycle time of the loop will result in some values being returned far more often than would otherwise be expected.
In the worst case, you end up with a stepped counter.

Code:

    for ( int i = 0 ; i < 100 ; i++ ) {
        printf("%d %ld\n", i, random_number(1000000,0));
    }

0 333106
1 335437
2 337008
3 338460
4 339900
5 341341
6 342776
7 344219
8 345655
9 347091
10 348527
11 350011
12 351456
13 352878
14 354304
15 355735
16 357165
17 358595
18 360030
19 361457
20 362885
21 364317
22 365749
23 367186
24 368617
25 370041
26 371467
27 372900
28 374324
29 375752
30 377176
31 378602
32 380025
33 381456
34 382887
35 384313
36 385739
37 387166
38 388590
39 390019
40 391447
you can guess how this keeps going up slowly....
90 510122
91 511548
92 512970
93 514396
94 515821
95 517640
96 523882
97 525437
98 526891
99 528325

Originally Posted by manpage

CLOCK_REALTIME
System-wide clock that measures real (i.e., wall-clock) time. Setting this clock requires appropriate privileges. This clock is affected by discontinuous jumps in the system time
(e.g., if the system administrator manually changes the clock), and by the incremental adjustments performed by adjtime(3) and NTP.

External agents can mess about with the clock, making your so-called random numbers even worse.

You're assuming that tv_nsec is actually as fast as it might suggest.
You need to call clock_getres() to test that assumption.

Famous Quote:
Randomness is too important to be left to chance.

**gaxio** · 04-20-2019

Code:

// Calculate the random number.
    number = tp.tv_nsec % max;

No. Very no. You should look up PRNG algorithms before trying to implement this. The LCG and XOR Shifter algorithms are very easy to implement and vastly superior to just getting the current time and dividing it.

**Yonut** · 04-21-2019

Originally Posted by Salem

Famous Quote:
Randomness is too important to be left to chance.

Ty for showing me that I wasn't testing with a large enough data set. You really showed the lack of randomness. This explains why I have not seen this method used.

Originally Posted by gaxio

No. Very no. You should look up PRNG algorithms before trying to implement this. The LCG and XOR Shifter algorithms are very easy to implement and vastly superior to just getting the current time and dividing it.

Ty for giving me a direction in which to learn more. This is why I wanted to post my code.

Thread: My random number generator. (Looking for feedback)

Thread Tools

Search Thread

Display

My random number generator. (Looking for feedback)

Similar Threads

Feedback on prime number generator source code?

Random Number Generator Stuck on 1 Number?

Only getting 0; random number generator

need a random number generator thats not compleatly random

random number generator

Tags for this Thread