Friday, 18 September 2009

Long runs of primes without a full period prime

Having had an on-and-off interest in primes for a very long time, I have recently come to learn a great deal more about prime numbers in recent years.
One topic that I have become curious about is the periods of various recurring decimals. Recurring decimals like 0.2727272727272727272727272727272727.... were a topic extensively covered in school mathematics, but those with periods longer than two were never discussed, so that a fraction like 0.142857142857142857142857142857...... remained unknown to me until I looked seriously at primes on the web.

Since that time, I have learned a great deal about recurring decimals that I was never taught in schools. The most basic thing to learn is:
  1. the period of reciprocal of every prime number p divides into p-1
  2. that if the remainder of p when divided by 40 is 1, 3, 9, 13, 27, 31, 37 or 39
  3. then p-1 divide by the period of reciprocal will be even
  4. whilst if the remainder of p when divided by 40 is 7, 11, 17, 19, 21, 23, 29 or 33
  5. then p-1 divide by the period of reciprocal will be odd
  6. a prime with period p-1 is a full period prime
It is known that something like three-eighths of primes will be full-period primes, though for certain sequence of primes the proportion is zero (above) whilst for others it is very high. For instance:
  • Of the primes in the sequence 23, 143, 263, 383, 503, 623, 743, 863 (red numbers are composite)
  • the first prime that is not full-period is 20903, which has period of reciprocal 2986.
In recent times I have been curious as to what the longest sequence of primes without a full-period prime is. I have known of the sequence from 3037 to 3121 and the earlier one from 751 to 809 because of the short-period prime 3061 which is the number of wickets taken by the great bowler John Thomas Hearne. However, I had no idea about larger sequences until recently, when because I found it so strange that there were no full-period primes among those eleven.

I will now list sequences of primes not full-period. In red are primes that are not full-period by virtue of the remainder when divided by 40.
  • 3037, 3041, 3049, 3061, 3067, 3079, 3083, 3089, 3109, 3119, 3121
  • (11 primes)
  • 37483, 37489, 37493, 37501, 37507, 37511, 37517, 37529, 37537, 37547, 37549, 37561, 37567, 37571, 37573
  • (15 primes)
  • 84871, 84913, 84919, 84947, 84961, 84967, 84977, 84979, 84991, 85009, 85021, 85027, 85037, 85049, 85061, 85081
  • (16 primes)
  • 104707, 104711, 104717, 104723, 104729, 104743, 104759, 104761, 104773, 104779, 104789, 104801, 104803, 104827, 104831, 104849, 104851
  • (17 primes)
  • 113921, 113933, 113947, 113957, 113963, 113969, 113983, 113989, 114001, 114013, 114031, 114041, 114043, 114067, 114073, 114077, 114083, 114089
  • (18 primes, but only three could potentially have full period)
  • 133117, 133121, 133153, 133157, 133169, 133183, 133187, 133201, 133213, 133241, 133253, 133261, 133271, 133277, 133279, 133283, 133303, 133319, 133321, 133327
  • (20 primes, but only five could potentially be full-period)
  • 461413, 461437, 461441, 461443, 461467, 461479, 461507, 461521, 461561, 461569, 461581, 461599, 461603, 461609, 461627, 461639, 461653, 461677, 461687, 461689, 461693, 461707, 461717, 461801, 461803
  • (25 primes - note the two-prime century - of which only four could be potentially full-period)

No comments: