Dualism

[robert44444uk]

Gentlemen

In order to bring a bit of life into this thread, I am wondering if we should not concentrate our efforts on those Sierpinski k which have neither the form k.5^n+1 nor k+5^n prime. The second form is the "dual" and share exactly the same covering sets (or not!) as their proth equivalent.

I have run the remaining Sierpinski candidates through as dual numbers and found prps for all but five candidates, tested up to n=28000. The five remaining are:

31712 37292 93254 96994 109988

I am also trying the same for the Riesel candidates, but it is harder, as the values of -k+5^n, for small powers of n, are negative, and therefore must be discarded and there were a lot more candidates to begin with. Right now at n=20000 I still have 39 candidates. I won't post them just yet.

Regards

Robert Smith

[fetofs]

Originally Posted by robert44444uk

]

I have run the remaining Sierpinski candidates through as dual numbers and found prps for all but five candidates, tested up to n=28000. The five remaining are:

31712 37292 93254 96994 109988

I've raised the upper limit of 31712 to 29500. Is working together (all n) or separately better?

[robert44444uk]

I think it is probably best to sieve at 28000 because -f100 or similar tool in pfgw does not go deep enough, and is inefficient. So take one candidate at a time, and run the sieve which takes you to 100000 or so.

Good luck!

Robert Smith

[fetofs]

Originally Posted by fetofs

I've raised the upper limit of 31712 to 29500. Is working together (all n) or separately better?

I've raised the lower bound to min n untested=42281. Going to stop here, I think (maybe not, who knows)