# Finding prime numbers code issue

Show 40 post(s) from this thread on one page
Page 1 of 2 12 Last
• 06-26-2013, 01:55 AM
Devin_p
Finding prime numbers code issue
I'm a beginner in JS and can't figure out why this isn't working right.

It prints out "2,2,2,2,2,2,2" and so on, but I need it to print the first 100 prime numbers.

Any help is appreciated.

Code:

``` var primeNumbers = []; //Prime numbers stored here while(primeNumbers.length <= 100){                 //Number to be tested if it is a prime.         var candidate = 2;         for(var i = 2; primeNumbers.length <= 100; i++)         {                 if(candidate % i == 0){                         continue;                 }else{                         i = 0;                         primeNumbers.push(candidate);                 }         };                 candidate++; }; for(var print = 0; print <= primeNumbers.length; print++){         //Prints out the prime #'s stored in var primeNumbers.                 console.log(primeNumbers[i] + ","); };```
• 06-26-2013, 02:31 AM
Old Pedant
Code:

```for(var print = 0; print <= primeNumbers.length; print++){         //Prints out the prime #'s stored in var primeNumbers.                 console.log(primeNumbers[i] + ","); };```
• 06-26-2013, 02:32 AM
Old Pedant
But the algorithm you are using still won't work.

It's horribly broken.
• 06-26-2013, 02:42 AM
Devin_p
Thank you, i didn't catch that, but it is still printing 2 over and over.

For a more efficient way to find primes, do you mean the Sieve of Eratosthenes?
• 06-26-2013, 03:27 AM
Devin_p

Changed it a bit, although still bad and broken..

Code:

``` var primeNumbers = []; //Prime numbers stored here while(primeNumbers.length <= 100){                 //Number to be tested if it is a prime.         var candidate = 2;         for(var i = 2; i <= Math.sqrt(candidate); i++)         {                 if(candidate % i === 0){                         continue;                 }else{                         primeNumbers.push(candidate);                         i = 2;                 }         };                 candidate++; }; var output = primeNumbers.join(); //Combines array into string and prints it. console.log(output);```
• 06-26-2013, 04:06 AM
Old Pedant
The Sieve is the most efficient that I know of.

But even one like yours can be improved two-fold by just considering odd numbers.
• 06-26-2013, 04:26 AM
Old Pedant
But even better: Why try moduloing the candidate with *every* number (even every odd number)?? Why not just try testing it against ALL THE PRIMES LESS THAN itself!
Code:

```<script type="text/javsascript"> // Prime numbers stored here // ..start with 2 as only even prime var primeNumbers = [2]; var candidate = 3;  // start with first odd number while ( primeNumbers.length < 100 ) {     var prime = true; // assume candidate is prime     // no point in checking against primeNumbers[0],     // which is 2.  Because our candidate value will     // never be even...so start at element 1 of the array instead     for ( var i = 1; i < primeNumbers.length; ++i )     {         if ( candidate % primeNumbers[i] == 0)         {             prime = false;             break; // no point in continuint         }     }     if ( prime ) primeNumbers.push(candidate);     candidate += 2; // skip the even numbers }; alert( primeNumbers ); </script>```
• 06-26-2013, 04:49 AM
Old Pedant
Even getting all the prime numbers less than 1000, the Sieve method is roughly 3 times faster than the method shown above. I would expect its advantage to grow as the number of prime numbers desired grows.

EDIT: I couldn't stand it; had to test my gut feelings.

Yes, if you ask for all prime numbers less than 5000, then the Sieve outperforms the modulo by a factor of *TEN*.

Not surprising, but I just wanted to check my sanity.
• 06-26-2013, 05:51 AM
Devin_p
Thank you for all your help and effort you put into this! :thumbsup:
• 06-26-2013, 05:58 PM
Old Pedant
Just for completeness, here's the version of the Seive of Erastothenes that I used:
Code:

```<script type="text/javascript"> var nums = [ ]; var primes = [ ]; var MAXNUM = 5000; // change this as desired for ( var n = 0; n < MAXNUM; ++n ) {     nums[n] = n; } // run the actual seive for ( var chk = 2; chk < MAXNUM; ++chk ) {     if ( nums[chk] != 0 )     {       primes.push(chk);       for ( var mark = chk; mark < MAXNUM; mark+=chk )       {           nums[mark] = 0;       }     } } alert(primes); </script>```
There are ways to make that a little more efficient, but not enormously so.
• 06-26-2013, 06:30 PM

Quote:

Originally Posted by Old Pedant
But even better: Why try moduloing the candidate with *every* number (even every odd number)?? Why not just try testing it against ALL THE PRIMES LESS THAN itself!

Even better: Only try all integers less than or equal to the square root of the number [1][2] :) And now combining these two solutions, even better is to check all prime numbers less than or equal to the number [3].

As for algorithms in general: First off, the Sieve of Eratosthenes could be improved, but one could also use another optimized version known as the Sieve of Atkin which is better by a factor of 1/(log log n).

[1] If p was not prime, it could be factored into p = ab. Without loss of generality, assume a > sqrt(p). It would immediately follow that b < sqrt(p). Since finding only one factor suffices to disprove that p is a prime, it is enough to check all integers less than or equal to the square root of the number in question.

[2] The number of primes <= x is asymptotically equivalent to x/ln(x). The limit of that divided by sqrt(x) is -- as one can see with fairly basic maths -- infinite, which means that it is by far better to use sqrt(x). As an additional benefit the test is independent of all prime numbers smaller than the number in question.

[3] Again, if p was not prime, there would be a factorization p = ab. In [1] we have seen that we can assume a <= sqrt(p). Now, if a was not prime itself, it could be further factorized into a = xy. The same argument would show (without loss of generality) that x <= a <= sqrt(p).
• 06-26-2013, 06:38 PM
DrDOS
Quote:

Originally Posted by Old Pedant
But even better: Why try moduloing the candidate with *every* number (even every odd number)?? Why not just try testing it against ALL THE PRIMES LESS THAN itself!

Actually, it only needs to be test against all primes less than the square root of itself, so you just need to make a list of them.
• 06-26-2013, 07:52 PM
Old Pedant
Well, even the version of the Sieve of Erastothenes that I used is not the best, as I said. I know it can easily be made about 4 times as efficient. But that code is pretty simple and compact, and unless you are going for a huge number of primes it's more than adequate.
• 06-27-2013, 07:10 AM
@DrDos
Less than or equal to(!) -- this is crucial, otherwise you would count every square of a prime number as prime.

@OldPedant

I agree that there is no need to improve it if it satisfies your needs. But switching to the square root is not a factor of four, we're talking orders of magnitude and a legit asymptotic benefit here. It's not a very special improvement either, it's fairly common to write even simple implementations using the square root as the bound. So yes, for the first few hundred primes it should be practically irrelevant, but I believe that with technical algorithms like this, at least showing what impact little details can have is important. :)
• 06-27-2013, 06:41 PM
Old Pedant
Okay, I give up. How does using the square root of the number being checked apply to the Sieve of Erastothenes algorithm???

Look at my code in post #19.

To do the prime.push(chk), we must loop all the way to the max num we are wanting to check.

And then the inner loop must also go all the way to that same max num to be sure all the multiples of the given prime are marked.

But notice that the marking loop at least only starts at the found prime number, so it's not terrible. But we could get 4 times as efficient by only considering odd numbers and by not bother with marking even multiples of a prime number. Oh, and we don't need to mark the prime number we just found.

Further than that...I dunno. What do you think can be done? And how does square root apply to this algorithm??
Show 40 post(s) from this thread on one page
Page 1 of 2 12 Last