Run P-1 before PRP tests

[alpertron]

Originally Posted by axn

http://www.mersenne.ca/prob.php?expo...tton=Calculate says 5.68%. For 42 tests, that amounts to 2.5 factors. You found 7. Sure that's a lot more, but not impossibly so. The problem is low sample size. Maybe once you've run it for about 1000 exponents, we can have some degree of confidence on the numbers.

I finally found 33 factors in the range of exponents from 2300000 to 2310000. The number of Mersenne number processed was 465. So I found one factor for each 14 numbers.

M2300017, M2300057, M2300153, M2300563, M2300731, M2300999, M2301199, M2301251, M2301421, M2302889, M2303137, M2303239, M2303321, M2303867, M2304839, M2304961, M2306597, M2306957, M2307059, M2307247, M2307451, M2307941, M2308181, M2308391, M2308483, M2308637, M2308919, M2309323 (two prime factors), M2309689, M2309693 (two prime factors), M2309771.

[axn]

Originally Posted by alpertron

I finally found 33 factors in the range of exponents from 2300000 to 2310000. The number of Mersenne number processed was 465. So I found one factor for each 14 numbers.

M2300017, M2300057, M2300153, M2300563, M2300731, M2300999, M2301199, M2301251, M2301421, M2302889, M2303137, M2303239, M2303321, M2303867, M2304839, M2304961, M2306597, M2306957, M2307059, M2307247, M2307451, M2307941, M2308181, M2308391, M2308483, M2308637, M2308919, M2309323 (two prime factors), M2309689, M2309693 (two prime factors), M2309771.

So 33/465 = 7.1%. Let's say the true probability is between 6.9-7.3%. In that case, the calculated probability of 5.7% is an underestimate, but not a huge one (about 20% off). Of course, I still would like to see a rigorous calculation of the proper probability.

[jocelynl1204]

Originally Posted by VictordeHolland

If anybody wants my P95 P-1 savefiles (B=10e6, B2=200e6) from exponents in the range 1.5M-1.7M (on 2500+ exponents without factors) to extend them for instance, let me know.

Note that the exponents in question also have had 280 ECM curves @B1=50,000 done since the P-1 runs. Somebody better educated in the subject can probably calculate if it still is useful to run more P-1 or just ECM.

Yes Vicotor I want them. And if anybody wich to send me their P-1 saved files in range 1M-4M. I would be more than happy to download them.