• 0

[C] random double between 0 and 1


Question

Hello all,

Im a newbie to C and I am not figuring out how to make a function that will generate a double between 0 and 1...

I KNOW NOTHING ABOUT C but managed to get a hello world work...

I tries this function but i think it's not working properly:

float_number = (float) rand()/RAND_MAX;

as when i try to display the num with this forloop it's getting me huge negative or positive numbers

		for(i = 0; i < 500; i++){
			printf("%d\n",  float_number);
		}

another thing, what should i add to the code so i dont allow the program to exit directly after execution and so the command prompt stays opened so i can see the results?

Thanks in advance.

Elio

Link to comment
https://www.neowin.net/forum/topic/980266-c-random-double-between-0-and-1/
Share on other sites

21 answers to this question

Recommended Posts

  • 0

You should use "%f" for floating point numbers, not "%d."

On the second question, some IDEs will prevent the program from closing immediately. Otherwise something simple like fgetc(stdin) will work.

Thanks for the fgetc(stdin) it worked...

the random number gives always the same number :s

think i passed through this earlier today on google.. will double check...

10x anw

  • 0

That is because by default the rand function is seeded with 0. Check this out: http://www.cplusplus.com/reference/clibrary/cstdlib/srand/ the most popular solution is to seed it with the system time. That should give you different pseudo random sequences for each run.

  • 0

for(i = 0; i < 500000; i++){
			double_number = (double) rand()/RAND_MAX;
				srand (time(NULL)); // this is used to seed the random number by the system time
			printf("%f\n",  double_number);
		}

The thing is i need to generate a huge number of random doubles in a for loop...

i hate 2 problems:

  1. Time interval in the loop is verry small, the number is slightly changing every one second which is bad
  2. The generated numbers are very closed to each other,,, and i dont want that :(

OKay fixed it :)

\

for(i = 0; i < 50; i++){
			double_number = (double) rand()/RAND_MAX;
				srand (time(NULL)+rand()); // this is used to seed the random number by the system time
			printf("%f\n",  double_number);
		}

  • 0

Call srand(time(NULL)) once before the loop. This will continue to generate random numbers. You don't need to call it every time.

Also, since you're using double precision instead of single, you really should use "%Lf" in the printf statement.

Okay thanks...

It's actually my first C program :-)

  • 0

I KNOW NOTHING ABOUT C

Cool. So maybe one of the first things that you should learn is that rand() does not, under any stretch of your imagination, generate a random number. Pound that idea into your head. It doesn't matter what you seed it with, either, but seeding with the current time is what everybody does and it almost always causes a problem at some point. Someone could use a number of methods to predict, both ahead of time and after the fact, the exact 50 doubles that your function would generate. If you run your program more than once within 1 second, it will generate the same sequence of 50 doubles. Or if two people run your program within a second of each other they will both generate the same sequence. If you need a range of doubles other than 0.0 to 1.0, I guarantee you'll do it wrong. Your rand() function will start to repeat itself after about 32,000 calls. There are plenty of other pitfalls. These things may or may not matter to you, but the reality is that this kind of thing is a bit complicated.

There is no good cross-platform way to reliably generate random numbers without using C++ or using a C library. This is especially true if you're looking for something like numbers that are unpredictable, has a special distribution (normal, uniform, poisson, etc), want them in a range that's not evenly divisible by the size of the random numbers you are generating, or if you need a secure seed value.

If you're able to use C++ (which will compile C code as well), modern compilers support a set of random number generators that can generate random numbers in various distributions (random, normal distribution, even distribution, etc) using various algorithms (including unpredictable sources), with larger periods (the amount of times before the RNG starts repeating). That way you don't have to worry about clipping your range or accidentally creating an uneven probability for certain numbers (like lots of people do with "rand() % SOME_NUMBER"). For example, you can do this to reliably generate a random (meaning each double has an equal probability of being generated) double between 0.0 and 1.0, inclusive:

mt19937 engine(YOUR_SEED);
uniform_real<double> distribution(0.0, 1.0);
variate_generator<mt19937&, uniform_real<double> > gen(engine, distribution);

for (int i = 0; i < 50; i++)
    double rndDouble = gen();

You can generate more values than there are atoms in the visible universe before that generator repeats itself. Then getting other ranges, distributions, generators, or engines is as easy as switching them in.

Regardless of whether you can use C++, if you need a seed value, then please seed your rand function with something other than time(NULL); you can find countless forum posts where using time-based seeds causes some problem. On linux, just read a random value from "/dev/urandom". On Windows, you can use rand_s (which makes a call to the cryptographic RNG function, which is stupidly-named "SystemFunction036") if speed is a concern or use "rand_s" directly and simply skip the seeding process (since a cryptographic RNG, by definition, doesn't need a user-supplied seed value).

  • Like 2
  • 0

another thing, what should i add to the code so i dont allow the program to exit directly after execution and so the command prompt stays opened so i can see the results?

you can use

system("PAUSE");

which is a call to the PAUSE instruction you can also call in cmd

  • 0

Okay thanks everyone...

I managed to make the assignment work...

it is part of the parallel programming course and we are trying to calculate the value of PI based on the mote carlo (somthing like that :p lol ) method...

and i need to take 100% random numbers.

The thing is we did it sequentially just to get used to C as we are JAVAists

and the next step would be parallelizing it...

thanks again

  • 0

//check if a point is inside a circle
		for( i=0; i<number_of_points; i++){   //number_of_points = 99999999;
			double current_x = points_xes[i];
			double current_y = points_yies[i];    //is this causing huge memory usage?


			if((current_x * current_x) + (current_y * current_y) <= 1){
				count_inside++;
			}else{
				count_outside++;
			}

I THINK THIS PART OF CODE IS USING MY RAMS... AFTER EXECUTION THE RAM USAGE INCREASES BY 1 GB... do you think this loop is the reason? as i have huge number of points and i'm creating 2 new doubles in each iteration?

how can i fix this?

Should i set them to null in the end of the loop and later "garbage collect them"?? (is there garbage collection in c?)

thanks

  • 0

If you're storing almost 100 million doubles in memory, yes that will tend to use up a lot of memory at at least 8 bytes a pop.

There is garbage collection in C, but it's manual. You use the free(my_ptr) function where my_ptr is something that was created using malloc or calloc (like the huge arrays you're using). You don't perform garbage collection on regular doubles. It also shouldn't be recreating them, but you can move the double current_x to oustide the loop and then just do current_x = if somehow that's the case.

  • 0

You are not creating two new doubles each iteration. You are re-using the same doubles that live on the stack.

However, assuming the code works correctly, points_xes and points_yies contain at least 100 million elements. Assuming they are doubles, and a double is 8 bytes, that is 2 x 8 bytes x 100 million = 1600MB. The memory allocation happens before the loop, in some code you haven't shown here.

  • 0

After seeing that 100,000,000 iteration loop, I'm not sure we want to see the other code. I also don't want to know if he's stack-allocating 100,000,000-length double arrays.

#include<stdio.h>
#include<stdlib.h> //now i can use rand()

#include<time.h>
#include<string.h>


int i;
double double_number;
int number_of_points = 99999999;
double points_xes[99999999];
double points_yies[99999999];
int count_inside = 0; //count the number of points inside the circle
int count_outside = 0; //count the number of points outside the circle
double percentage_p_inside = 0; //percentate of the points inside the circle
double PI = 0; //the found value of PI


time_t start_time;
time_t end_time;




main()
{

		printf("Welcome to computing pi!, Elio's first C program.");
		printf("\n\nComputing PI based on %d",number_of_points);
		printf(" points.\n");

		start_time = time (NULL); //get the start time

		srand (time(NULL)+rand()); // this is used to seed the random number by the system time
		//computing the first number_of_points point and considering them as the "X" value for each point
		for(i = 0; i < number_of_points; i++){
			double_number = (double) rand()/RAND_MAX;

			points_xes[i] = double_number;
			//printf("%Lf\n",  double_number);
		}


		//computing the second number_of_points point and considering them as the "Y" value for each point
		for(i = 0; i < number_of_points; i++){
			double_number = (double) rand()/RAND_MAX;

			points_yies[i] = double_number;
			//printf("%Lf\n",  double_number);
		}



		//check if a point is inside a circle
		for( i=0; i<number_of_points; i++){
			double current_x = points_xes[i];
			double current_y = points_yies[i];


			if((current_x * current_x) + (current_y * current_y) <= 1){
				count_inside++;
			}else{
				count_outside++;
			}
		}



		printf("\nNumber of points inside: %d",count_inside);
		printf("\nNumber of points outside: %d", count_outside);

		//calculate the percentage of points inside the circle
		percentage_p_inside = (double)(count_inside)/(count_inside + count_outside);
		printf("\nPercentage of points inside the circle: %Lf",percentage_p_inside);


		PI = percentage_p_inside * 4;
		printf("\n\nPI = %Lf", PI);

		end_time = time(NULL);

		printf("\nPI calculated in: %d",(end_time-start_time));
		printf(" seconds.");

		fgetc(stdin); // just to avoid closing the command prompt after the excecution of the code.
}


here is the rest of the code...

I know it could be very bad in term of design but i told u it's my first c program and i done it all by myself.. they didnt give us any tutorial

it's just a university small hw... :p

10x

238423149.jpg

  • 0

I'm surprised the program even compiles or runs, although I've never tried putting that much data in static storage so I don't really know. Maybe someone with better knowledge of C can provide some precisions. Anyway, when you need large amounts of memory, use the heap (malloc() in C, new in C++). This will allow you to free it at runtime when you are done with it (free() in C, delete in C++), whereas static storage remains until program termination. See here for examples (in C++).

That being said, even using dynamic memory allocation isn't the way to go. Your basic problem here is that you try to generate all your points first, and then perform your statistical analysis on the whole bunch. What you could do, and it would use almost no memory, is to only work with a single point at a time, i.e. (pseudocode):

int pointsInsideCount = 0;
int pointsOutsideCount = 0;

for (number of points) { 
    double x = generateRandomDouble();
    double y = generateRandomDouble();
    if ((x, y) is inside circle) {
        ++pointsInsideCount;
    }
    else {
        ++pointsOutsideCount;
    }
}

This only ever uses the same two doubles, on the stack. 8 temporary bytes vs 1600 static MBs. Plus here it doesn't matter how many points you want, it'll take longer but it'll never use up any additional memory. :yes:

  • 0

I'm surprised the program even compiles or runs, although I've never tried putting that much data in static storage so I don't really know. Maybe someone with better knowledge of C can provide some precisions. Anyway, when you need large amounts of memory, use the heap (malloc() in C, new in C++). This will allow you to free it at runtime when you are done with it (free() in C, delete in C++), whereas static storage remains until program termination. See herefor examples (in C++).

That being said, even using dynamic memory allocation isn't the way to go. Your basic problem here is that you try to generate all your points first, and then perform your statistical analysis on the whole bunch. What you could do, and it would use almost no memory, is to only work with a single point at a time, i.e. (pseudocode):

int pointsInsideCount = 0;
int pointsOutsideCount = 0;

for (number of points) { 
    double x = generateRandomDouble();
    double y = generateRandomDouble();
    if ((x, y) is inside circle) {
        ++pointsInsideCount;
    }
    else {
        ++pointsOutsideCount;
    }
}

This only ever uses the same two doubles, on the stack.

yes u're right...

wat a bad design i was using :s loool

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Posts

    • But the reality is it will work for people's needs, and they don't care about the technology that makes it. Clearly not everyone's needs, but that low end space where personal laptops were only used to type emails, watch content and browse websites, but they didn't want to do that on a small screen device. Heck, writing that out I can now see the connection and reason it'll do so well. Apple is about experience. If the experience is bad, they don't release it. Low end Windows laptop manufacturers up until this point have not taken that into consideration ever before, so slow laggy usage with brittle slimey plastic shells were common. I hope that the low end space at least creates better physical products that last a bit longer, and if Microsoft get their act together, they could also have a solid OS on such low end hardware that would actually make the experience work for what the hardware was intended for. The fact that the CPU is a "cellphone", sorry mobile phone processor is irrelevant. It's about the experience, and so far, that sounds quite solid.
    • Hello, Bonjour is Apple's implementation of a multicast-DNS service, which allows devices running Apple's software and/or hardware to find each other on your local network.  I believe the Windows version was last updated around 2010. If you do not need it, you can stop and disable the Bonjour service in the Services Control Manager (filename: SERVICES.MSC).  Once you have done that, the operating system will no longer attempt to load the service. Regards, Aryeh Goretsky  
    • This AMD RX 9070 16GB GPU that performs close to Nvidia 5070 is under $600 by Sayan Sen With the memory shortage that's prevalent nowadays, discounts are super-hard to get. As such we post good deals whenever they pop up. Recently, we covered a few great discounts on SSDs wherein you can get a 4TB TeamGroup NVMe PCIe Gen4 drive for just $400 thanks to a special coupon. If you want a faster product but don't need all that capacity, you can also opt for Samsung's 990 PRO 2TB that is on sale for its lowest price in over three months. Let's say though that you are on the hunt for a 1440p gaming card. In that case AMD's RX 9070 non-XT can help, and with its 16GB VRAM, you can also run AI models locally without worrying about bottlenecking (check out our recent 9070 GRE reviews for gaming and productivity to get an idea). The PowerColor Reaper variant of the RX 9070 is currently on sale for just $580 which is a very good price in the current state of affairs (purchase link under the specs table down below). The Reaper cooler on this 9070 uses a triple‑fan design with ring‑blade fans, paired with premium dual ball bearings to extend lifespan and reduce friction. "Intelligent" fan control allows the fans to remain idle at lower temperatures, only spinning up when the GPU is under load. A nickel‑plated copper base makes direct contact with both the GPU and memory modules, helping to spread heat evenly. PowerColor also applies Honeywell PTM7950 phase‑change thermal interface material (TIM), which fills microscopic gaps between the die and heatsink for more efficient thermal transfer. The fan shroud is shorter in height as the firm has made it such that it can be used in certain SFF (small form factor) cases. The technical specifications of the Reaper RX 9070 are given in the table below: Specification Value Stream Processors 3584 Units Video Memory 16GB GDDR6 Memory Speed 20.0 Gbps Memory Interface 256-bit Engine Clock Game Clock: up to 2070 MHz Boost Clock: up to 2520 MHz Bus Standard PCI Express 5.0 x16 Display Connectors 1 x HDMI 2.1b, 3 x DisplayPort 2.1a Maximum Resolution DisplayPort: 7680 × 4320 HDMI: 7680 × 4320 Board Dimensions 289mm × 111mm × 41mm 304mm × 127mm × 42mm (with bracket) Slot 2 Minimum System Power Requirement 600W Power Connectors Two 8-pin PCI Express Get the PowerColor Reaper RX 9070 at the links below (you get only a 90-day warranty on Woot): PowerColor Reaper Radeon RX 9070 16GB Graphics Card (RX9070 16G-A): $579.99 (Sold and Shipped by Amazon US) (Was: $700) PowerColor Reaper Radeon RX 9070 16GB Graphics Card (RX9070 16G-A): $559.99 (Sold and Shipped by Woot US) Good to know This Amazon deal is U.S. specific, and not available in other regions unless specified. We only use first-party seller links (at the time of article publishing); ensure that you purchase from a first-party seller link only. Check out Today's Deals on Amazon | or our recent tech deals. Become a Prime member (for Students or SNAP) via Neowin Get Prime Access - Prime for half price (for qualifying Medicaid, EBT, SNAP) Subscribe to Prime Video, Audible Plus, Music Unlimited or Kindle Unlimited via Neowin As an Amazon Associate, we earn from qualifying purchases.
    • Are they marketed as an entry into astronomy or astrophotography? I do astrophotography. With big rigs, lots of computers, cables and headaches. I love it. And by learning this ridiculously complex hobby, I’ve learned about the objects I’m shooting. Astronomy followed from photography.
    • Microsoft confirms Recycle Bin bug across all versions of Windows by Usama Jawad A couple of days ago, we reported that the latest Patch Tuesday update has seemingly resulted in a lot of issues for many users, including OneDrive and Dropbox access problems, BitLocker recovery lockouts, and BSODs. Although Microsoft is yet to acknowledge these bugs, it has confirmed another, relatively smaller issue across all supported versions of Windows. In an update on its Windows Release Health Dashboard, Microsoft has confirmed that after installing June's Patch Tuesday update (KB5094126), you'll experience unexpected behavior when leveraging Recycle Bin. Basically, when you attempt to delete an item from the Recycle Bin, the confirm dialog will show you the internal file name of that content rather than the actual name. For example, the file may be named abc.png, but the confirm dialog will ask if you're sure that you want to permanently delete $Rxxxxx.png from the Recycle Bin. This is pretty much it for the scope of the bug itself; it just displays the wrong name in the confirm dialog. The correct name will be shown in the list view of the Recycle Bin and if you restore the file, it will return with the correct name as well. This issue affects pretty much all supported versions of Windows client and server, including: Client: Windows 11, version 26H1; Windows 11, version 25H2; Windows 11, version 24H2; Windows 11, version 23H2; Windows 10, version 22H2; Windows 10 Enterprise LTSC 2021; Windows 10 Enterprise LTSC 2019; Windows 10 Enterprise LTSB 2016 Server: Windows Server 2025; Windows Server 2022; Windows Server 2019; Windows Server 2016; Windows Server 2012 R2; Windows Server 2012 As things currently stand, Microsoft is working on a concrete solution that will be released in a "future" Windows update. It remains to be seen if the firm will wait till the next Patch Tuesday or roll out an out-of-band (OOB) fix. The good news is that commercial customers can deploy a workaround right now, but they will have to reach out to Microsoft Support for Business for additional details.
  • Recent Achievements

    • Week One Done
      Jordan Smith earned a badge
      Week One Done
    • Reacting Well
      BizSAR earned a badge
      Reacting Well
    • First Post
      AndreaB earned a badge
      First Post
    • Week One Done
      Huge Trailer earned a badge
      Week One Done
    • Week One Done
      Classifyskilleducation earned a badge
      Week One Done
  • Popular Contributors

    1. 1
      +primortal
      579
    2. 2
      +Edouard
      184
    3. 3
      PsYcHoKiLLa
      75
    4. 4
      Michael Scrip
      72
    5. 5
      neufuse
      64
  • Tell a friend

    Love Neowin? Tell a friend!