greedy computation of pseudoperfect numbers?
T. D. Noe
noe at sspectra.com
Tue Apr 8 17:33:23 CEST 2008
Can someone prove or find a counterexample
to the potential comment for A136446 --
``odd members are A005231'' ?? R.
On Tue, 8 Apr 2008, T. D. Noe wrote:
> At 8:45 AM -0400 4/8/08, Maximilian Hasler wrote:
>>> For both sorts of pseudoperfect numbers, allowing divisor one A005835,
>>> and forbidding divisor one A136446, a greedy method _seems_ to always
>>> work for deciding whther a given n in in the sequence:
>
> Let d be the list of divisors of n (we can include or exclude 1). Compute
> terms of the polynomial product_i (1+x^d[i]) up to the x^n term. (Perhaps
> your PARI code does this. I don't read PARI.) If the x^n term is nonzero,
> then n is the sum of a subset of the divisors in d. This is fairly fast
> and deterministic. Using this method, the first 10 odd terms of A136446 are
>
> 945, 1575, 2205, 2835, 3465, 4095, 4725, 5355, 5775, 5985
>
> which starts the same as A005231, odd abundant numbers. Here is the
> Mathematica code:
>
> t = {}; n = 1; While[Length[t] < 10, n = n + 2;
> d = Rest[Most[Divisors[n]]];
> c = SeriesCoefficient[
> Series[Product[1 + x^d[[i]], {i, Length[d]}], {x, 0, n}], n];
> If[c > 0, AppendTo[t, n]]]; t
>
> Best regards,
>
> Tony
>
>
>
More information about the SeqFan
mailing list