[seqfan] Re: nice new board game puzzle A337663

[hv at crypt.org]

Peter Kagey <peterkagey at gmail.com> wrote:
:It looks like Arnauld removed his post for some reason. However, there is a new program that was posted that computes a(5) = 49: https://codegolf.stackexchange.com/a/212182/53884 <https://codegolf.stackexchange.com/a/212182/53884>

I can confirm my own code also finds a(5) = 49.

Altogether, I have:
a(2) = 16 (0.00s, 523 positions traversed)
a(3) = 28 (0.16s, 195,551 positions)
a(4) = 38 (44s, 63,200,517 positions)
a(5) = 49 (14054s, 18,371,175,865 positions)

(Running times on my computer.)

Based on those figures I expect it would take 30-60 CPU days to find
a(6) (my guess is about 35), but due to hardware problems I have no
chance of keeping the computer up that long.

The big difficulty is verifying correctness, so I plan to work some
more on writing tests for my code and looking for further speed
improvements.

If somebody fancies running my code or looking at it, it's available
at https://github.com/hvds/seq/tree/master/A337663 - let me know if
you run into any problems.

Hugo