12
$\begingroup$

The complementary Bell numbers or Uppuluri–Carpenter numbers, denoted $\tilde{B}_n$, can be delivered by $$G(x):=\sum_{n\geq0}\tilde{B}_n\frac{x^n}{n!}=e^{1-e^x}.$$ Definition. Fix an integer $m\geq0$. We say $F(x)=\sum_{n\geq0}a_n\frac{x^n}{n!}$ has the index-m property whenever $a_n=0$ iff $n=m$ is satisfied.

In this language, Herb Wilf's conjecture can be stated as: $G(x)$ has the index-2 property.

By now, there are a few papers turning this into a theorem, including this one:

Tewodros Amdeberhan, Valerio De Angelis and Victor H. Moll, Complementary Bell Numbers: Arithmetical Properties and Wilf’s Conjecture (2013) In: Kotsireas I., Zima E. (eds) Advances in Combinatorics. Springer, Berlin, Heidelberg, doi:10.1007/978-3-642-30979-3_2 (pdf)

The other day, I started pondering on a truncated version of the generating function, namely, for $\pmb{r\geq2}$, define $$G_r(x)=e^{-(x+\frac{x^2}{2!}+\frac{x^3}{3!}+\dots+\frac{x^r}{r!})}=\sum_{n\geq0}\tilde{B}_{n,r}\frac{x^n}{n!}.$$ Of course, $\lim_{r\rightarrow\infty}G_r(x)=G(x)$. After some experimentation, I'm enough convinced to ask:

Question 1. Fix $r\geq2$. Is it true that $G_r(x)$ has the index-2 property?

Question 2. Does Wilf's conjecture imply the claim in Question 1? Or, vise-versa?

An aside: the functions $G_r(x)$ are equally interesting as related to partition theory, Hermite polynomials, permutations, Young tableaux, etc.

UPDATE.

(1) From $G_{r+1}(x)=G_r(x)\cdot e^{-\frac{x^{r+1}}{(r+1)!}}$, it follows that $\tilde{B}_{n,r+1}=\tilde{B}_{n,r}$ for all $n\leq r$. By the same reasoning, $\tilde{B}_n=\tilde{B}_{n,r}$ for all $n\leq r$. It is now clear that $\tilde{B}_{n,r}=0$ iff $n=2$, for all $r$, implies $\tilde{B}_n=0$ iff $n=2$. This related to a part of Question 2.

(2) It is also possible to prove if $G(x)$ has index-2 property then so does $G_r(x)$, for large enough $r$. Again, this is related to a part of Question 2.

(3) The case $G_2(x)$ has index-2 property is known:

Solution by R. Israel to a problem proposed by F. Schmidt and R. Simion, Even minus odd involutions in the symmetric group, SIAM Rev. 34 (2), pages 315-317 (1992) doi:10.1137/1034062.

$\endgroup$
2
  • $\begingroup$ mathoverflow.net/questions/261409/… $\endgroup$ Commented May 30, 2017 at 14:46
  • $\begingroup$ Israel's explicit formula implies that the (positive) binomial transform of $\tilde B_{n,2}$ is an aerated sequence of double factorials. The binomial transform for general $r$ experimentally has small prime factors, suggesting a similar explicit formula. $\endgroup$
    – MTyson
    Commented Aug 3, 2018 at 1:10

0

You must log in to answer this question.