wheel factorization

To see if a number is prime via trial division (or to find its prime factors), we divide by all of the primes less than (or equal to) its square root.  Rather than divide by just the primes, it is sometimes more practical to divide by 2, 3, and 5; then divide by all the numbers congruent to 1, 7, 11, 13, 17, 19, 23, and 29 modulo 30--again stopping when you reach the square root. These are the positive integers less than, and relatively prime to, the primes 2, 3 and 5.

This type of factorization is called wheel factorization and the spokes are the list of integers prime to all of the primes we are using.  To see if 3331 is prime using the wheel mentioned above, we would divide by 2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 49, and 53.  Notice that 49 is not prime, but is relatively prime to the spokes of the wheel.

Wheel factorization is special type of sieve.

Wheels can be made of any size. The simplest wheel would be created by just using the single prime 2, and there would be one spoke: 1 (so we would divide by 2, then each of the odd integers).  A little more complicated wheel would use the primes 2 and 3. This would give us two spokes: 1 and 5. So we would divide by 2, 3 then the following integers:

5, 7, 11, 13, 17, 19, 23, 25, 29, 31, 35, 37, 41, 43, ...

This is the same as writing the integers in a table of width 6--all of the primes (other than 2 or 3) end up in the columns corresponding to the two spokes 1 and 5.  (The primes are colored green.)

123 456
789101112
131415161718
192021222324
25 2627282930
31323334 3536
373839404142

If we start with the primes 2, 3, 5, and 7; then we must check the numbers modulo 210 (the product of these four primes) which are congruent to the spokes:

1, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59, 61, 67, 71, 73, 79, 83, 89, 97, 101, 103, 107, 109, 113, 121, 127, 131, 137, 139, 143, 149, 151, 157, 163, 167, 169, 173, 179, 181, 187, 191, 193, 197, 199, and 209.

Notice that other than 1, the smallest spokes in these wheels are all prime (the composites are colored red).  This is because any number less than the square of the largest prime used in the wheel, not removed by the wheel, must be prime.  This has misled many to suppose that such wheels (and patterns) are good "generators" of primes.  They are not.  The density of primes decreases as the integers increase in size (see the prime number theorem), so when we apply these same wheels to a list of large integers, almost all of those that are not removed by the wheel are composite.

The simple wheel based on 2 and 3 removed 4/6 = 2/3 rds of the composites.  The larger wheel using 2, 3, 5, and 7 will remove 162/210 (over 77%) of the composites.

Large wheels are not particularly efficient.  To remove 90% of the composites, we must use the primes up to 251.  95% requires the primes to 75,037.  96% requires the primes to 1,246,379.  97% requires the primes to 134,253,593!  Just imagine how many primes you will need to remove 99% of the composites!

See Also: TrialDivision, SieveOfEratosthenes

Printed from the PrimePages <t5k.org> © Reginald McLean.