Thread: k=13236795
View Single Post
Old 2021-04-03, 10:55   #90
gd_barnes's Avatar
May 2007
Kansas; USA

37·281 Posts

Some bad news...

I have completed a double-check up to n=1M on this k. I found 7 missing primes. They are:


It appears that some sieving was done incorrectly for the range of n=220K-450K. I looked at the statuses in the old 15k search thread and found that your number of tests was much less than were in a file sieved by me to P=50T.

Here is the thread:

Here is a comparison of your number of tests in that thread compared to the number of tests in a file sieved to P=50T:

post #11; n=220K-231K; 744 tests; sieve 1439 tests; 51.7% tested
post #20; n=231K-240K; 576 tests; sieve 1183 tests; 48.7% tested
post #21; n=240K-250K; 674 tests; sieve 1319 tests; 51.1% tested
post #22; n=250K-280K; 1997 tests; sieve 3944 tests; 50.6% tested
post #23; n=280K-300K; 1295 tests; sieve 2652 tests; 48.8% tested

Summary total for n=220K-300K:
5286 tests; sieve 10537 tests; 50.2% tested; 5251 tests missing

Your sieve depth back then was far less than P=50T so you should have more than 100% tested of my file.

Clearly this only covers part of the "problem" 220K-450K range but the above was all that I found with enough detail on your number of tests to make a comparison to my file.

I suspect that post #6 was the error. Where you state "Currently resieving the range 220000 through 500000 with Proth 0.42...found 3000+ factors which NewPGen initially not found". You may have been sieving for Proth primes, i.e. +1 instead of Riesel -1, and the program removed a lot of good Riesel tests. Also 3000+ factors after you have already sieved for quite awhile with NewPGen seems like a large number. I suspect that was in error.

It appears that you later used a different sieve for n=450K-1M so the range of n=450K-500K looks OK.

I also looked at your results for n=1M to 1.3M. The number of tests keeps rising abnormally as the testing gets higher:
n=1.0M-1.1M 12752 tests
n=1.1M-1.2M 13218 tests
n=1.2M-1.3M 14251 tests

n=1.2M-1.3M has 11.8% more tests than n=1.0M-1.1M, which is an unusual fluctuation considering it is a higher n-value. You may want to look into that. For comparison a sieve to P=50T has 13293, 13210, and 13139 tests respectively for the ranges.

I will continue double-checking this k up to your current test limit of n=1.3M. It should be done in ~5 days.

I've attached the primes and residues for n=0-1M and the sieve file for n=0-2M that is sieved to P=50T.

The good news is that this k now has 113 primes for n<=1.3M...a very large number. :-)

Attached Files
File Type: zip (2.47 MB, 41 views)
gd_barnes is offline   Reply With Quote