[seqfan] Re: Finite sequence?

[M. F. Hasler]

If you have enough memory to memoize, i.e., store at least the "nontrivial"
values of A008284(n,k),
(say, 3 < k < n/2: for larger k it reduces to A41(n-k) and for k=1,2,3 you
have 1, [n/2], [n^2/12+.5]),
then I think this should not take a week to calculate: since T(n,k) = T(n-k,k)
+ T(n-1,k-1),
you need less than n/2 additions, each requiring 2 table lookups, to
calculate a (half-)row,
which makes less than 5000^2 additions (admittedly, of big integers) to
store,
plus the 5000 values of A41 (which my slow laptop calculates in about 2
sec).

- Maximilian

On Tue, Sep 26, 2017 at 9:48 AM, Hans Havermann <gladhobo at bell.net> wrote:

> > Blindly using the second (of three) Mathematica formulation in A008284,
> I'd guess (based on a half-day run) that it would take me about three days.
>
> It ended up taking more than a week! Jonathan, If you want to check it
> against your 3-minute output, I'll put my A008284(10000) here:
>
> http://chesswanks.com/num/a008284(10000).txt