Formula for A120014(n), Anyone?
Paul D. Hanna
pauldhanna at juno.com
Thu Jun 8 07:06:01 CEST 2006
Max (and Seqfans),
Interleaving my replies below ...
On Wed, 7 Jun 2006 20:23:29 -0700 Max <maxale at gmail.com> writes:
>
> If you bring C_k as C(2k,k)/(k+1) it is natural to split the
> formula
> into three parts:
> n^2/(n-1))^(n+1) / n^4 * (2*n-1)
> (n^2/(n-1))^(n+1) / n^4 * (2-n)/n * Sum_{k=0..n-1} ((n-1)/n^2)^k
C(2k,k)
> - (n^2/(n-1))^(n+1) / n^4 * 2 * Sum_{k=0..n-1} ((n-1)/n^2)^k
C(2k,k)/(k+1)
This is helpful. Yet some identity is needed to simplify.
> These sums can give also an approximation for a(n) as n tends to
infinity.
Good insight.
> > Although this is not any simpler, the presence of the factorials
> suggest
> > that simplification may require some identity of LambertW.
>
> Could you please explain this point (maybe with a smaller example)?
> What is the relationship between factorials and Lambert W-function?
Yes, my wording was vague at best.
What I was refering to here was the following lovely
LambertW convolution identity:
Sum_{k=0..n} p*(n-k+p)^(n-k-1)* q*(k+q)^(k-1)* C(n,k)
= (p+q)*(n+p+q)^(n-1)
which is the coefficient of x^n/n! in:
W(x)^(p+q) = W(x)^p * W(x)^q
where W(x) = -LambertW(-x)/x.
My objective was to restate your formula into a form
where this identity may be applied.
We may restate your 3 parts into a convolution form:
a(n) = (n^2/(n-1))^(n+1)/n^4*(2*n-1)
+(n^2/(n-1))^2/n^4*(2-n)/n * Sum_{k=0..n-1} (n^2/(n-1))^(n-1-k) *
C(2*k,k))
-(n^2/(n-1))^2/n^4*2 * Sum_{k=0..n-1} (n^2/(n-1))^(n-1-k) *
C(2*k,k)/(k+1))
and then the sums include factors like (n-1)^(n-1-k) ...
which reminded me of LambertW.
But LambertW may not be involved at all -
I will have to check.
Thanks,
Paul
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.seqfan.eu/pipermail/seqfan/attachments/20060608/b3688ea4/attachment-0001.htm>
More information about the SeqFan
mailing list