Analyzing table of optimal B1 in GMP-ECM's README file. The sequence is neither linear nor geometric. It is roughly quadratic. b1_ainc.c
(as of 6.3-5ubuntu1) needs updating.
Min | N | Max | Linear | Geometric | Log log |
---|---|---|---|---|---|
11e3 | 74 | 5e4 | 527.027027 | 0.142856 | 2.456124 |
5e4 | 214 | 25e4 | 934.579439 | 0.0570377 | 2.274716 |
25e4 | 430 | 1e6 | 1744.186047 | 0.0314601 | 2.230921 |
1e6 | 904 | 3e6 | 2212.389381 | 0.0160494 | 2.131485 |
3e6 | 2350 | 11e6 | 3404.255319 | 0.00676381 | 2.047746 |
11e6 | 4480 | 43e6 | 7142.857143 | 0.00385742 | 2.055486 |
43e6 | 7553 | 11e7 | 8870.647425 | 0.00238583 | 2.018008 |
11e7 | 17769 | 26e7 | 8441.668074 | 0.00105949 | 1.923939 |
26e7 | 42017 | 85e7 | 14041.935407 | 0.000480654 | 1.897046 |
$d=$b-$a; $s=sprintf "<tr><td>%s %d %s %f %g %f</td></tr>", $a, $n, $b, $d/$n, log($d)/$n, log($d)/log($n); $s=~s, ,</td><td>,g
No comments :
Post a Comment