Skip to main content
12 events
when toggle format what by license comment
Sep 4, 2013 at 19:12 history edited Pietro Majer CC BY-SA 3.0
added 25 characters in body
Aug 13, 2013 at 8:28 history edited Pietro Majer CC BY-SA 3.0
format
Nov 16, 2010 at 10:27 history edited Pietro Majer CC BY-SA 2.5
added 1 characters in body
Nov 15, 2010 at 2:09 comment added Steffen Marcus Interesting... I was initially expecting the solution to be a generating function argument of some type, so your approach was a pleasant surprise. I've been unsuccessful in trying to naively write the generating function for either $A(n)$ and $B(n)$ as a convolution. In fact, they originate from playing with polynomials in the Lambert function.
Nov 15, 2010 at 0:24 comment added Pietro Majer You're welcome! I think you can get further information on the sequence $B(n)$ via a GF. One should start by a GF for the $U_n(x)$, say $u(x,t):=\sum_n U_n(x) t^n$ or $u(x,t):=\sum_n U_n(x) t^n/n!$ that satisfies a PDE coming from the recurrence of the $U_n$ (or another convenient variation). Then $b(t):=\int_{-\infty}^0 u(x,t)x/(1-x) dx$ is essentially a GF for the $B(n)$.
Nov 14, 2010 at 23:51 comment added Steffen Marcus Thanks Pietro! This is very slick and just the sort of thing we've been hoping for... certainly a much more interesting argument than I expected. Thanks also to Mike and Martin for your efforts/feedback.
Nov 14, 2010 at 23:34 vote accept Steffen Marcus
Nov 14, 2010 at 19:33 history edited Pietro Majer CC BY-SA 2.5
added 42 characters in body
Nov 14, 2010 at 19:01 history edited Pietro Majer CC BY-SA 2.5
added 196 characters in body
Nov 13, 2010 at 23:03 history edited Pietro Majer CC BY-SA 2.5
added 4 characters in body
Nov 13, 2010 at 22:57 history edited Pietro Majer CC BY-SA 2.5
added 234 characters in body; added 3 characters in body; edited body
Nov 13, 2010 at 22:47 history answered Pietro Majer CC BY-SA 2.5