[seqfan] Re: Binary Complement Sequences

[Tim Peters]

[Tom Duff <eigenvectors at gmail.com>]

YAY!!!! So all seeds <= a million are known to hit 0 eventually.

> I have them all:
> 819991 87037147311
> 820560 25996066240
>
> 829974 4327838170
>
> 836375 4327838183
>
> 836666 1212877574
>
> 875089 1246707601
>
> 876382 1246707550
>
> 879731 1246707639
>
> 973544 1246707556
>
> 977281 1246707553
>
> 980258 1246707530

I didn't report this before, but the last 6 are deeply related. In fact,
they're in a group of 21 that all hit 6,625,680 after no more than 155
iterations. Here are all 21:

[204953, 213884, 225095, 373290, 406923,
 407354,433716, 433870, 554427, 744312,
 787446, 795541, 796003, 807222, 807968,
 875089, 876382, 879731, 973544, 977281,
 980258]

> I haven't actually finished computing 819991, but it is known to be one
> less than 425720. I'm letting it run as a check on 425720.

Smart! It's annoying that there's no clear way (that I've thought of) to
run a sanity check in parallel as the computation goes along.

Alas, my computer did reboot after a power glitch (storm related), so I
have to start over on 425720 :-( But at least this time it's running newer
code that saves a checkpoint to disk every 100 million steps.