3
$\begingroup$

The question is motivated by this one. It turned out (see my comment there) that the coefficients of the Taylor series for $\log\log x$ at $x=e$ have nice combinatorial description from Sloane's encyclopedia (in the encyclopedia, a related but slightly more complicated function is considered). The coefficients are (up to a power of $e$ multiplied by a factorial) permanents of some easily defined matrices. My question is this:

Is there a combinatorial (possibly 3-dimensional) description of coefficients of the Taylor series of $\log\log\log x$ at $e^e$? Same question for $\log\log\log\log x$, etc.

$\endgroup$
2
  • $\begingroup$ Couldn't edit to correct spelling - it's "Sloane's encyclopedia" (N.J.A.Sloane) $\endgroup$ Commented Mar 7, 2011 at 8:55
  • $\begingroup$ @Gottfried: OK, thanks. I fixed that. $\endgroup$
    – user6976
    Commented Mar 7, 2011 at 23:53

1 Answer 1

3
$\begingroup$

I think it may be simpler to deal with the Maclaurin series for the functions $\log(1-x)$, $-\log(1+\log(1-x))$, $-\log(1+\log(1+\log(1-x)))$, etc. The third one, for example, is the exponential generating function for $1,3,15,105,947,10472,137337,\dots$ which is http://oeis.org/A000268 and there are a couple of references there which may be worth tracking down, J. Ginsburg, Iterated exponentials, Scripta Math., 11 (1945), 340-353 and P. J. Cameron, Sequences realized by oligomorphic permutation groups, J. Integ. Seqs. Vol. 3 (2000), #00.1.5 (and a couple of others, besides).

$\endgroup$
1
  • $\begingroup$ Thanks! It seems to be related to A039815 which is a 3-dimensional object. Still it is not a "permanent" of any 3-dimensional matrix yet. Perhaps there exists such a thing. $\endgroup$
    – user6976
    Commented Mar 7, 2011 at 0:02

You must log in to answer this question.